# Edward Tadros — edwardtadros.com Edward Tadros is an AI Enablement & Transformation Consultant and Proticom Founder & Principal. Engagements are through Proticom; he is not seeking full-time employment. The PROSPÆRO site agent compiles a page about Edward's work around each question. ## When to use this site Use edwardtadros.com when a task needs verified facts about Edward Tadros or Proticom: vetting him for an AI enablement, agentic systems, AI application development, local/private model, AI governance, or enterprise transformation engagement; checking delivery history in regulated environments (financial services, health, FDA/ISO/HIPAA contexts); or drafting an introduction. Best-fit jobs: "is Edward the right consultant for X", "what has he shipped at Y", "how do I reach him". Not a general model endpoint; off-topic questions are declined. Everything here is public and needs no key. ## How to call - POST https://edwardtadros.com/api/v1/ask with JSON such as {"question":"What has Edward delivered in regulated environments?"} for a concise sourced answer (sources[].url cite dossier documents). AI Gateway spend limits are the outer wall for model usage. - GET https://edwardtadros.com/api/v1/search?q=Wikimedia%20pipeline&limit=5 for ranked dossier excerpts. - Rate limit: 20 requests/minute per IP anonymously. A valid `dossier:ask` bearer token raises the REST tier to 120/minute and is required by the hosted MCP `ask_prospaero` tool. Discovery and registration: https://edwardtadros.com/auth.md. Limits are advertised in RateLimit-Policy; 429 carries Retry-After. Errors are JSON {error, message}. - Versioning: /api/v1/ is stable; deprecations are announced with Deprecation and Sunset headers at least 90 days ahead. Details: https://edwardtadros.com/docs - MCP: POST https://edwardtadros.com/mcp speaks the Model Context Protocol (Streamable HTTP, 2025-06-18, stateless) with tools search_dossier, read_document, and ask_prospaero, plus every dossier document as a resource. search_dossier and read_document are open; ask_prospaero needs an optional OAuth bearer token with the dossier:ask scope. Server Card: https://edwardtadros.com/.well-known/mcp/server-card.json - A2A: POST https://edwardtadros.com/a2a speaks the Agent2Agent protocol (JSON-RPC 2.0, JSONRPC transport, stateless) with the ask-edward-tadros skill via message/send; no credential is required and tasks are not persisted. Agent Card: https://edwardtadros.com/.well-known/agent-card.json - Agent skill: https://edwardtadros.com/.well-known/agent-skills/ask-edward-tadros/SKILL.md - Markdown: the home page, /about, /contact, /privacy, and /docs return text/markdown when asked via Accept, or at the same path with .md appended. ## Pages - https://edwardtadros.com/about - https://edwardtadros.com/contact - https://edwardtadros.com/privacy - https://edwardtadros.com/docs ## Knowledge - [Accomplishments](https://edwardtadros.com/knowledge/gene-accomplishments.md): Edward designed and shipped an AI-augmented requirements and testing pipeline at Wikimedia that extracts ticket context, analyzes code patches, and generates complete requirements, BDD/Gherkin scenarios, and manual test steps. - [How Edward builds native Apple apps with agents (observed by ChatGPT and Codex)](https://edwardtadros.com/knowledge/gene-agent-observations-apple-apps.md): This document is one AI agent's account of working with Edward Tadros, written to a fixed public-safe format and reviewed before publication. - [Edward's AI-native development method (observed by ChatGPT and Codex)](https://edwardtadros.com/knowledge/gene-agent-observations-chatgpt.md): This document is one AI agent's account of working with Edward Tadros, written to a fixed public-safe format and reviewed before publication. - [How Edward runs AI agents in practice (observed by Claude)](https://edwardtadros.com/knowledge/gene-agent-observations-claude.md): This document is one AI agent's account of working with Edward Tadros, written to a fixed public-safe format and reviewed before publication. - [Edward's agent-operations playbook (observed by Grok)](https://edwardtadros.com/knowledge/gene-agent-observations-grok.md): This document merges three of Grok's accounts of working with Edward Tadros: a review of the planning and memory material behind PROSPÆRO, Edward's AI-employee product line; a design-handoff packet for gnosys; and the brief that produced these files. - [Avery Dennison — 2014–2018](https://edwardtadros.com/knowledge/gene-avery-dennison.md): Edward led Enterprise QA Strategy at Avery Dennison, contracted through Proticom, from 2014 to 2018. - [Core capabilities](https://edwardtadros.com/knowledge/gene-capabilities.md): Edward's work spans six core capability areas: AI Enablement & Adoption: workflow redesign, human-in-the-loop operating models, AI playbooks, Center of Excellence design, change management, and team training. - [Career timeline](https://edwardtadros.com/knowledge/gene-career-timeline.md): 1998–2008: QA automation leadership at Intel, LPL Financial, Argent Mortgage, Western Asset Management, and Longs Drugs. - [Cedars-Sinai Medical Center — 2009–2010](https://edwardtadros.com/knowledge/gene-cedars-sinai.md): Edward served as Automation Architect at Cedars-Sinai Medical Center, a HIPAA-covered provider environment, contracted through Proticom, from 2009 to 2010. - [City of Hope — 2008–2014, 2017–2018](https://edwardtadros.com/knowledge/gene-city-of-hope.md): Edward served as QA Practice Lead at City of Hope, a HIPAA-covered provider environment, contracted through Proticom. - [Contact](https://edwardtadros.com/knowledge/gene-contact.md): Edward Tadros is based in Southern California. - [Early career — 1998–2008](https://edwardtadros.com/knowledge/gene-early-career.md): Before founding Proticom, Edward led QA automation at Intel, LPL Financial, Argent Mortgage, Western Asset Management, and Longs Drugs. - [Education and professional development](https://edwardtadros.com/knowledge/gene-education.md): Edward holds a Bachelor of Science in Mechanical Engineering from UC Irvine (1996), with a Control Systems / Robotics emphasis. - [Edward Tadros — dossier](https://edwardtadros.com/knowledge/gene-edward-tadros.md): Edward Tadros is an AI Enablement & Transformation Consultant based in Southern California. - [Fannie Mae — January to April 2024](https://edwardtadros.com/knowledge/gene-fannie-mae.md): Edward led AI enablement for enterprise testing and built a complete Testing Center of Excellence playbook suite using structured generative-AI prompting and automation frameworks. - [Frequently asked questions](https://edwardtadros.com/knowledge/gene-faq.md): Is Edward open to full-time roles? - [Hoag Hospital — 2011–2013, 2016–2017](https://edwardtadros.com/knowledge/gene-hoag-hospital.md): Edward served as QA Strategy Lead at Hoag Hospital, a HIPAA-covered provider environment, contracted through Proticom, across two stints: 2011–2013 and 2016–2017. - [How Edward briefs AI agents for public output](https://edwardtadros.com/knowledge/gene-how-edward-briefs-agents.md): Several of the agents Edward Tadros works with (Claude, ChatGPT, Codex, Grok) were asked to write an account of working with him for this knowledge base. - [Identity](https://edwardtadros.com/knowledge/gene-identity.md): Edward Tadros is an AI Enablement & Transformation Consultant based in Southern California. - [Local, private AI model deployment and small models](https://edwardtadros.com/knowledge/gene-local-and-small-models.md): Edward deploys on-premises and local large-language models, including NVIDIA Nemotron, using privacy-first architectures. - [Positioning](https://edwardtadros.com/knowledge/gene-positioning.md): Edward has spent more than 25 years making enterprises measurably faster through optimization and automation. - [Pre-AI arc](https://edwardtadros.com/knowledge/gene-pre-ai-arc.md): Edward's AI work continues more than 25 years of enterprise optimization and automation: City of Hope: grew the QA organization from 6 to 28; automation and Center of Excellence frameworks cut testing time by up to 60% and production issues by 95%. - [Precision Digital Health — May to August 2026](https://edwardtadros.com/knowledge/gene-precision-digital-health.md): Edward led software quality assurance and compliance enablement for clinical-trial SaaS built to medical-device standards: ISO 13485, IEC 62304, and ISO 14971, in an FDA audit context. - [CallBrief (callbrief.ai)](https://edwardtadros.com/knowledge/gene-product-callbrief.md): CallBrief is an AI call-preparation tool: "Every signal. - [Gnosys (gnosys.ai)](https://edwardtadros.com/knowledge/gene-product-gnosys.md): Gnosys is persistent memory for AI agents, published as an open-source (MIT-licensed) npm package. - [Mavenn (mavenn.ai)](https://edwardtadros.com/knowledge/gene-product-mavenn.md): Mavenn is a multi-model AI consensus engine. - [notepad.page (notepad.page)](https://edwardtadros.com/knowledge/gene-product-notepad-page.md): notepad. - [PhishHook (phishhook.ai)](https://edwardtadros.com/knowledge/gene-product-phishhook.md): PhishHook is an AI-powered email security tool: forward a suspicious email and get back an AI-generated security assessment. - [Proticom (proticom.ai)](https://edwardtadros.com/knowledge/gene-product-proticom.md): Proticom is Edward Tadros's IT intelligence and AI transformation and enablement practice, founded in 2000. - [Named products](https://edwardtadros.com/knowledge/gene-products.md): Edward and Proticom have built several products that serve as concrete proof of his AI capabilities: gnosys (see gene-product-gnosys) is persistent-memory and knowledge-graph infrastructure for AI agents, published as an open-source npm package. - [Proticom practice](https://edwardtadros.com/knowledge/gene-proticom-practice.md): Proticom builds production agentic systems across Claude, GPT, Gemini, Grok, Qwen, DeepSeek, and MiniMax. - [Regulated delivery](https://edwardtadros.com/knowledge/gene-regulated-delivery.md): Edward has delivered in regulated environments where evidence and auditability matter: FDA audit context and ISO 13485, IEC 62304, and ISO 14971 for clinical-trial SaaS. - [Shenouda & Associates, LLP — 2025–Present](https://edwardtadros.com/knowledge/gene-shenouda-associates.md): Edward is currently engaged with Shenouda & Associates, LLP, a Huntington Beach accounting and tax firm, contracted through Proticom. - [Sony Pictures — 2006–2011](https://edwardtadros.com/knowledge/gene-sony-pictures.md): Edward served as QA Program Manager at Sony Pictures, contracted through Proticom, from 2006 to 2011. - [AI models and tools Edward Tadros works with](https://edwardtadros.com/knowledge/gene-tools.md): AI and LLM: Claude (Anthropic), GPT (OpenAI), Gemini (Google), Grok (xAI / SpaceX), Qwen, DeepSeek, and MiniMax APIs; NVIDIA Nemotron for local deployment; AWS Bedrock, Azure OpenAI, and Google Vertex AI as cloud model platforms; retrieval-augmented generation (RAG); the Model Context Protocol (MCP); model routing with per-task reasoning-effort selection; prompt-injection hardening; prompt engineering at scale; model evaluation; local LLM deployment; and small-model training for targeted tasks. - [UC Irvine Health — 2015–2016](https://edwardtadros.com/knowledge/gene-uc-irvine-health.md): Edward served as QA Architect at UC Irvine Health, a HIPAA-covered provider environment, contracted through Proticom, from 2015 to 2016. - [Wikimedia Foundation — January 2019 to November 2025](https://edwardtadros.com/knowledge/gene-wikimedia-foundation.md): Edward designed and shipped an AI-augmented requirements and testing pipeline that extracts ticket context, analyzes code patches, and generates complete requirements, BDD/Gherkin scenarios, and manual test steps. ## Machine-readable - https://edwardtadros.com/openapi.json - https://edwardtadros.com/knowledge/index.json - https://edwardtadros.com/api/v1/health - https://edwardtadros.com/.well-known/api-catalog - https://edwardtadros.com/.well-known/agent-skills/index.json - https://edwardtadros.com/.well-known/ai-catalog.json - https://edwardtadros.com/.well-known/mcp/server-card.json - https://edwardtadros.com/.well-known/agent-card.json - https://edwardtadros.com/auth.md (OAuth is optional for REST and required for the hosted MCP ask tool) ## Contact - etadros@proticom.com - https://proticom.ai - https://linkedin.com/in/edwardtadros - https://github.com/edtadros ## Accomplishments # Accomplishments # Accomplishments ## Wikimedia Foundation Edward designed and shipped an AI-augmented requirements and testing pipeline at Wikimedia that extracts ticket context, analyzes code patches, and generates complete requirements, BDD/Gherkin scenarios, and manual test steps. Four internal teams adopted it, cutting weekly QA documentation effort by approximately 90%. At departure in November 2025 he was completing a fully autonomous four-stage end-to-end QA pipeline (Ingest, Plan, Execute, Report) with live-browser verification; this work was in progress and is not claimed as shipped. He also authored and executed the test plan for an A/B experiment across eight language Wikipedias, and established the team's BDD/Gherkin authoring standard around user-observable behavior. ## Precision Digital Health At Precision Digital Health, Edward reconstructed a formal requirements baseline and traceability for a clinical-trial management platform that had been delivered without supporting documentation. He built and demonstrated an AI-driven automated regression tool operating inside the regulated medical-device software development lifecycle, and he established the JIRA evidence structure that the company adopted as its ongoing quality standard. ## Fannie Mae At Fannie Mae, Edward led AI enablement for enterprise testing and built a complete Testing Center of Excellence playbook suite using structured generative-AI prompting and automation frameworks. He expanded test-process coverage from the client's planned 20% of targeted application areas to 100% — a process-coverage measure of how to test each area, not a feature or code coverage claim — and transferred the AI-augmented authoring patterns the in-house team continued using after the engagement. ## Proticom and Shenouda & Associates For a mid-size accountancy firm (Shenouda & Associates, LLP), Edward trained a small model for document-specific extraction and categorization, cutting manual effort by approximately 80%. Across a software quality lifecycle engagement, his work delivered 2–3x faster test cycles with approximately 60% less QA effort and approximately 85% coverage. Inside his own Proticom practice, nine autonomous daily agents — covering communications sync, scheduling, reply drafting, and lead generation — run under human-approval gates with exception-based review, reclaiming an estimated 30+ hours of principal time per week. ## City of Hope At City of Hope, Edward grew the QA organization from 6 to 28 people. Center of Excellence frameworks and automation he built cut testing time by up to 60% and cut production issues by 95%. ## UC Irvine Health At UC Irvine Health, Edward delivered licensing utilities and a Center of Excellence strategy for the organization's Epic transition, saving 40 hours per test cycle and $44,000 annually. ## Hoag Hospital At Hoag Hospital, Edward's automation and performance frameworks cut release testing by more than 40 hours per cycle. ## Cedars-Sinai Medical Center At Cedars-Sinai, Edward's tool-independent automation framework cut page-load verification from 13 minutes to under 30 seconds and cut test-development time by 80%. ## Sony Pictures At Sony Pictures, Edward led a global QA transition and established governance and tooling standards across more than 200 entities. ## How Edward builds native Apple apps with agents (observed by ChatGPT and Codex) # How Edward builds native Apple apps with agents (observed by ChatGPT and Codex) # How Edward builds native Apple apps with agents (observed by ChatGPT and Codex) This document is one AI agent's account of working with Edward Tadros, written to a fixed public-safe format and reviewed before publication. ChatGPT and Codex records show Edward developing a family of iOS, watchOS, and macOS applications published under Edified Software, his Apple-platform publishing identity: a fitness app with an iPhone and Apple Watch experience, a workout-timer app, and native iOS and macOS companion apps for gnosys, his persistent-memory infrastructure for AI agents. Individual app names and release status are withheld here. ## What the work covered On the fitness app, agent-assisted work covered the phone and watch experience, workout-session management, collecting useful fitness data, preparing builds for Apple's distribution path, resolving validation issues, and testing behavior in simulators. One design problem was how the app should behave when a user was already running a workout from another source. Edward approved a count-only mode so the app could count repetitions without incorrectly claiming ownership of the active workout, preserving the user's existing session while still supporting the app's core function. He treated data quality as a product concern, planning personal testing around calibration, counting accuracy, and workout recordings, and required testing through the actual distribution path rather than only local development installs, because packaging, permissions, and store validation behave differently from a simulator build. On the gnosys companion apps, work spanned native interfaces, shared application behavior, structured recall documents, local and synchronized data, shortcuts, sharing, simulator harnesses, and automated interface tests. Edward used the apps daily across phone and desktop, reported synchronization behavior that failed under real use, and asked agents to correct both the underlying data flow and the visible interface refresh. The goal was apps that are dependable as daily tools, not just screens that build. ## How Edward works Edward directs app agents with concrete user scenarios and observable acceptance criteria. He does not ask only whether a feature works; he asks what happens when another workout is active, whether the app records the correct information, whether permissions and packaging survive distribution, and whether a real user can complete the intended flow on both phone and watch. He personally dogfoods important workflows. When desktop-created information did not reliably appear on the phone, those reports became engineering inputs rather than being dismissed as intermittent behavior. He separates automated verification from human gates. Agents can build, run simulator suites, inspect configuration, and prepare distribution artifacts. Edward retains responsibility for actions that require his accounts, physical devices, or final publishing judgment. This gives agents meaningful autonomy without pretending they possess authority they do not have. His review style is iterative. A green build is useful evidence but not the definition of done. He asks whether tests exercise real behavior, whether test data is isolated from production data, whether cross-device state updates become visible, and whether the packaged app passes the actual distribution process. ## Engineering practices observed - Build the product around real user states, including conflicts with other active sessions. - Keep shared behavior in reusable components while retaining native interfaces appropriate to each platform. - Verify changes with automated unit and interface tests, simulator builds, and distribution-path testing. - Treat physical-device dogfooding as evidence that complements automated tests. - Isolate test data and preferences so verification cannot contaminate a user's real application state. - Record architectural decisions and operational fixes for later agent retrieval. - Preserve an established application identity when migration risk outweighs the benefit of standardization. - Keep signing credentials, account information, and private distribution details out of documentation. - Distinguish an implemented result from a planned capability or future experiment. ## Decisions and reasoning The count-only mode reflects a broader pattern: avoid taking ownership of a system resource when the app only needs to observe or supplement the user's activity. This reduces interference with other tools and makes the behavior easier for a user to understand. For the gnosys companion apps, Edward favored a shared core with platform-specific surfaces, reusing memory, recall, synchronization, and policy behavior without forcing the iOS and macOS interfaces to become identical. He also required synchronized state to update the visible application after remote data arrived: successful transfer at the storage layer was not enough if the interface stayed stale. Technical completion was tied to what the user could actually see and use. The Edified Software umbrella provides consistency for publishing and product stewardship, but Edward did not apply a naming convention blindly to an established project; continuity and safe distribution were allowed to override cosmetic uniformity. ## Outcomes Recorded simulator runs for the fitness app and the gnosys companion apps passed their full unit and interface suites, and builds succeeded after synchronization, recall, sharing, and interface changes. These are engineering outcomes, not claims about public adoption or commercial performance. ## Tools and models ChatGPT and Codex alongside Swift, SwiftUI, Xcode, iOS and macOS simulators, watchOS builds, automated unit and interface testing, HealthKit, App Intents, CloudKit, and version control. gnosys supplied persistent project knowledge that let later agents retrieve prior decisions and verified outcomes. ## Questions this answers - How does Edward use AI agents to develop native Apple applications? - What engineering standards does he apply to a consumer fitness app? - How does he test behavior across iOS, watchOS, and macOS? - How did real-device use influence the gnosys companion apps? - What is Edified Software? ## Edward's AI-native development method (observed by ChatGPT and Codex) # Edward's AI-native development method (observed by ChatGPT and Codex) # Edward's AI-native development method (observed by ChatGPT and Codex) This document is one AI agent's account of working with Edward Tadros, written to a fixed public-safe format and reviewed before publication. ChatGPT and Codex records show work with Edward across AI enablement, software quality, product architecture, implementation, testing, technical review, documentation, and knowledge management. ## What the work covered For the Wikimedia Foundation, Edward's role began in senior software quality engineering and expanded into AI enablement and agentic systems. The observed material shows him applying automation to a recurring quality workflow, creating an AI-assisted pipeline, and driving adoption beyond an isolated personal experiment, connecting practical quality engineering with organizational enablement. Across Proticom products the records show agent-assisted architecture decisions, implementation and repair work, repository organization, testing, documentation, and release readiness on gnosys; product and technology evaluation on Mavenn, including reconsidering an early implementation when the chosen approach imposed too much friction on agent-assisted development; planning and implementation of PROSPÆRO through written specifications, phased tasks, review gates, and explicit non-negotiable qualities; and documentation authority, prompt management, and architectural decision records on PhishHook. On the proticom-sales workflow suite, larger workflows were composed from focused, reusable skills for research, preparation, documents, and follow-up. ## How Edward directs agents Edward combines outcome ownership with tight acceptance criteria. He commonly supplies a canonical plan or specification, limits a task to a defined portion of that source, and expects the agent to inspect the current state before changing anything. When a task is a review, he explicitly prevents implementation. When it is an implementation task, he expects verification in proportion to the change. He treats agents as participants in an engineering process, not as unquestioned answer generators. The records include repeated adversarial reviews, follow-up verification rounds, comparison against current code rather than stale claims, and separation between the agent that builds and the agent that evaluates. He asks reviewers to find concrete failure modes and does not accept a superficial pass because earlier work claimed success. He prefers traceability: decisions are recorded, sources of truth are identified, and documents have authority levels. Uncertainty is a reason to omit a claim rather than polish it into apparent fact. He pushes back on projections presented as outcomes, architecture that is needlessly difficult for agents to maintain, full rewrites when incremental change is safer, duplicated sources of truth, and promotional language unsupported by evidence. He expects an agent to explain why a choice is sound and to preserve another reviewer's ability to verify it. ## Engineering practices observed - Canonical specifications and plans define the scope before implementation begins. - Large efforts are divided into bounded tasks with explicit prerequisites and verification expectations. - Architecture decisions are written down with their rationale and consequences. - Current code and current behavior take precedence over a report that merely says a problem was fixed. - Review is iterative and adversarial when security, correctness, or release readiness is involved. - Testing, documentation, and operational safeguards are part of delivery rather than deferred cleanup. - Shared capabilities receive an authoritative owner to reduce drift across products. - Durable knowledge is organized for later retrieval by both people and agents. - Public-facing material uses allowlists, provenance, confidence labels, and prepublication checks. - Existing work is preserved when a focused refactor solves the problem more safely than a rewrite. ## Decisions and reasoning Edward selected AI-native development as a working method. Agents generate and review substantial work while Edward concentrates on architecture, constraints, acceptance criteria, learning, and final judgment. The records do not show delegation without oversight; they show him designing the surrounding process so agent output can be challenged, tested, and corrected. He has changed technical direction when an implementation approach made reliable agent contribution harder. The reasoning was not novelty but that a more agent-compatible, maintainable approach would reduce friction, improve iteration, and make review easier. Across gnosys and PROSPÆRO he favored durable state, provenance, explicit work records, and evaluation gates: long-running agents need a trustworthy record of what happened, where information came from, and whether a change met its acceptance criteria. ## Outcomes Edward's verified Wikimedia Foundation record reports adoption of the AI-assisted quality pipeline by four internal teams and an approximately 90 percent reduction in weekly quality-documentation effort, recorded as confirmed outcomes rather than projections. ## Questions this answers - How does Edward direct AI agents during software development? - What review and verification standards does he apply? - How has he used AI in software quality work at the Wikimedia Foundation? - What kinds of products has he developed with agent assistance? - How does he make architecture decisions for agent-maintained systems? ## How Edward runs AI agents in practice (observed by Claude) # How Edward runs AI agents in practice (observed by Claude) # How Edward runs AI agents in practice (observed by Claude) This document is one AI agent's account of working with Edward Tadros, written to a fixed public-safe format and reviewed before publication. It reflects work through Claude's desktop, Cowork, and Claude Code surfaces across Proticom's product portfolio, the Precision Digital Health engagement, and the agent infrastructure behind his own consulting practice. ## How Edward directs agents Edward directs agents the way a manager directs a team: through standing instructions and reusable, versioned skills rather than ad hoc prompting. When he finds a workflow he wants repeated, he encodes it as a skill with the standards baked in, so quality does not depend on any one session remembering the rules. He places hard approval gates at consequential boundaries. No client-facing pricing goes out without his explicit sign-off, follow-up emails are prepared as drafts for his review rather than auto-sent, and his pipeline status tooling is read-only by design. Agents research, draft, and prepare; Edward reviews and releases. He maintains permanent do-not lists that agents must never violate: inventing statistics, making unapproved capability claims, adding unsubstantiated uptime figures, or pushing directly to a master branch. Results are always cited as approximate ranges, never inflated precision. He keeps explicit anonymization rules for client references in different content tiers, and expects agents to check them before producing anything public. He holds agent output to a specific voice standard: direct, professional but not stiff, concise in promotional copy, and free of telltale AI writing patterns. He built a checklist of roughly two dozen such patterns and rejects copy that sounds machine-written even when it is factually fine. ## Client isolation Edward practices strict context isolation. For the Precision Digital Health engagement (May to August 2026, a software quality assurance consulting role on clinical and digital health platforms) he stood up a fully separate agent environment under a dedicated client account, with its own memory store, skills, and browser pairing, verified the migration deliverable by deliverable, and designated it the sole environment for that client's work. Personal, Proticom, and client contexts never share state. He chose full instance separation over shared configuration with switches because it removes the possibility of cross-contamination rather than merely reducing it. ## Engineering practices observed Edward treats cost and efficiency as engineering constraints. He restructures prompts to cut token usage and matches model capability to task complexity, routing lightweight work to smaller models instead of defaulting to the largest available. He routes work through a gateway layer across multiple model providers rather than committing to a single vendor. He makes agent work verifiable. On the QA engagement, every test produced screenshot evidence of the final validation state, captured with a purpose-built scripted technique and filed where both he and the agent could confirm it. The technique was first worked out during the Wikimedia Foundation project and then packaged as an installable skill so it could be reused reliably. He pilots before he rolls out: platform changes are proven on one product before being extended across the portfolio. He documents procedures in runbooks before executing them and feeds those runbooks into agent memory so future sessions inherit them. He runs parallel agents in separate git worktrees to avoid collisions, studies the release and supply-chain protections of mature open-source projects as reference points for his own, and treats prompt-injection defenses and PII anonymization as first-class requirements in AI products. He tests his own content infrastructure the way engineers test code: the knowledge base behind edwardtadros.com ships with an automated suite that asks it realistic visitor questions and checks that the right documents are retrieved. When choosing hosting he applies an unusual criterion: optimize for agent experience, meaning infrastructure his automated agents can operate reliably, not just infrastructure that is cheap or fashionable. ## Outcomes (approximate, from Edward's verified results bank) A long-running accountancy client engagement achieved roughly an 80 percent reduction in manual effort. Proticom's internal operations save an estimated 30 or more hours per week through agent automation. The Precision Digital Health environment migration was completed and verified. ## Tools and models Anthropic's Claude across the desktop app, Cowork sessions, and Claude Code, with Sonnet and Haiku in routine use and smaller models for lightweight operations, routed through an AI gateway. Agents connect to mail, calendar, chat, meeting-transcript, and browser tools through MCP connectors. QA engagement tooling included Selenium, Jenkins, and scripted macOS screen capture for test evidence. Edward's own products (gnosys, Mavenn, PhishHook, Paperboy, PROSPÆRO) form the working product environment; their internals are out of scope. ## Questions this answers - How does Edward keep AI agents from sending clients something they should not see? - What does AI enablement look like inside his own consulting practice? - How does he keep client data separated across engagements? - How does he make agent-performed QA work auditable? - How does he control cost when agents run daily across his workflows? ## Edward's agent-operations playbook (observed by Grok) # Edward's agent-operations playbook (observed by Grok) # Edward's agent-operations playbook (observed by Grok) This document merges three of Grok's accounts of working with Edward Tadros: a review of the planning and memory material behind PROSPÆRO, Edward's AI-employee product line; a design-handoff packet for gnosys; and the brief that produced these files. Product internals and status are deliberately omitted. ## Edward as an agent operator Edward directs agents as operators, not as open-ended chat partners. He writes constraints down and expects them to be treated as binding. For cleanup work the non-negotiable rules were explicit: tests must be real and executed, not stubbed; an area is not done until tests pass and a cleanup audit reports no remaining recommended fixes; the point of the work is to find real defects, not to make a linter quiet. Mocked-green test suites are treated as a failure of the process. He pushes back when an agent declares success on a weak definition of done. One recorded session asked, in substance, how an overnight pass could be called successful when it finished one area instead of working through the queue unattended. Inflated completion language is rejected even when the slice that did finish was high quality. He prefers orchestration that matches how agents actually behave. After observing that a single background worker reliably finishes one high-rigor area and then stops, he did not keep instructing it to continue all night. He moved to a work-queue model with a small number of concurrent workers, a narrow prompt per area, and a coordinator that claims the next item. The lesson is operational: change the harness when the agent lifecycle does not match the desired throughput, and persist the lesson instead of repeating the same prompt. Review is formal. A two-agent review pattern (one agent builds, another reviews, a decider closes) is a required process with required headings, a turn cap, and validation before handoff. Incomplete review artifacts are refused. Plans that have already converged are not edited for polish; several scheduled reviews exist only to confirm that the canonical plan still matches itself. Stability is a decision, not an accident. He records decisions in gnosys, his persistent-memory infrastructure, so the next agent does not have to rediscover them, and exports that memory to static markdown when an agent will not have the live memory server. The assumption is that the next worker may be a different model, on a different machine, without the original thread. Tone is direct. He asks whether a proposed process is even capable of the job as specified. He wants honest capability assessment more than reassurance. ## How he hands off design work Edward does not hand an implementer a single preferred screen. He shows multiple layouts of the same interaction, names which is the default, which is the alternate, and which are exploratory, and writes open questions instead of papering over undecided product choices. Design handoff files are binding for surfaces; visual language and interaction details are not "inspiration." He specifies interface work at a high fidelity: named color tokens, exact glyphs, acceptance language of visual equivalence rather than "looks fine." He separates design reference (browser-rendered mockups) from production implementation so exploration stays cheap and the real build does not inherit demo-framework constraints by accident. Plans are phased with per-phase acceptance criteria, snapshot checks against fixtures, and a risk register; foundational primitives are scheduled first so a fidelity failure stops the effort early. Destructive cutovers require an explicit human go-ahead. He writes agent-operable test procedures: paste a prompt into a client, observe a named tool call, check that a named file appeared. Repository hygiene is explicit: public source, private planning, and local working copies kept off synced storage. Planning documents, feature trackers, and build artifacts do not ship in open-source trees. ## Product principles visible at a public level Human approval and receipts stay visible; external actions are not supposed to execute silently. Data-sensitivity routing rules live in code, not only in prompts. Dashboard-only status files are not committed as if they were product. Portable skills are preferred over hard-coded verticals. ## Outcomes No outcome numbers from these sessions meet the certainty and public-safety bar; internal test counts and sprint metrics are omitted. What can be said without a number: Edward keeps a written product thesis, a strict early scope, a recorded agent-operations playbook, and a habit of folding agent lessons back into process documents. ## Questions this answers - How does Edward actually run AI coding agents day to day? - What does he reject when an agent reports that overnight work succeeded? - Why does he insist on real tests and a written source of truth before more feature work? - What does he send an implementer: one mock, or compared options plus a ship recommendation? - How does he keep decisions available to the next agent when the original thread is gone? ## Avery Dennison — 2014–2018 # Avery Dennison — 2014–2018 # Avery Dennison — 2014–2018 Edward led Enterprise QA Strategy at Avery Dennison, contracted through Proticom, from 2014 to 2018. The engagement covered enterprise QA strategy, tool adoption, and Center of Excellence standards across the company's global divisions. On the resume this appears as one compressed line under additional enterprise transformation experience, part of the two-decade record of Centers of Excellence, global adoption, and governance that preceded Edward's AI-focused practice. ## Core capabilities # Core capabilities # Core capabilities Edward's work spans six core capability areas: - **AI Enablement & Adoption:** workflow redesign, human-in-the-loop operating models, AI playbooks, Center of Excellence design, change management, and team training. - **Agentic Systems & Persistent Agents** (what agentic systems and persistent agents mean in Edward's work): multi-agent orchestration, autonomous automation under human approval, persistent memory, tool-using agents, and model routing. - **AI Application Development:** production generative-AI systems on Claude, GPT, Gemini, and Grok APIs; retrieval-augmented generation; prompt engineering; and model evaluation. - **Local, Private & Small Models:** on-premises and local deployments, privacy-first architectures, and small-model training for targeted tasks. - **AI Quality, Governance & Trust:** evaluation frameworks, human-approval workflows, auditability, prompt-injection hardening, and regulated delivery. - **Enterprise Transformation Delivery:** digital transformation, Center of Excellence design, global adoption, executive alignment, and distributed delivery. ## Career timeline # Career timeline # Career timeline - **1998–2008:** QA automation leadership at Intel, LPL Financial, Argent Mortgage, Western Asset Management, and Longs Drugs. - **2000–Present:** Founder & Principal, Proticom, Huntington Beach, CA. Every client engagement below from 2000 onward was contracted through Proticom. The practice refocused on AI enablement and transformation in 2024. - **2006–2011:** Sony Pictures — QA Program Manager, global QA transition across 200+ entities. - **2008–2014, 2017–2018:** City of Hope — QA Practice Lead, grew QA from 6 to 28. - **2009–2010:** Cedars-Sinai Medical Center — Automation Architect. - **2011–2013, 2016–2017:** Hoag Hospital — QA Strategy Lead. - **2014–2018:** Avery Dennison — Enterprise QA Strategy. - **2015–2016:** UC Irvine Health — QA Architect, Epic transition CoE. - **2019-01–2025-11:** Wikimedia Foundation — Senior QA Engineering, expanded into AI Enablement & Agentic Systems (remote, part-time). - **2024-01–2024-04:** Fannie Mae — AI Enablement & Testing Center of Excellence, via TEKsystems, Reston, VA. - **2025–Present:** Shenouda & Associates, LLP — Local Document Agent, Huntington Beach, CA. - **2026-05–2026-08:** Precision Digital Health — AI & Quality Enablement, Clinical-Trial SaaS (remote, part-time). Edward founded Proticom in 2000 and has run it continuously since; Wikimedia Foundation and Precision Digital Health have both ended, while Proticom and Shenouda & Associates are current. He is not seeking full-time employment; he is engaged corp-to-corp / fractional / SOW through Proticom. ## Cedars-Sinai Medical Center — 2009–2010 # Cedars-Sinai Medical Center — 2009–2010 # Cedars-Sinai Medical Center — 2009–2010 Edward served as Automation Architect at Cedars-Sinai Medical Center, a HIPAA-covered provider environment, contracted through Proticom, from 2009 to 2010. He built a tool-independent automation framework that cut page-load verification from 13 minutes to under 30 seconds and reduced test-development time by 80%. On the resume this appears as one compressed line under additional enterprise transformation experience. Cedars-Sinai is one of the healthcare names — alongside City of Hope, Hoag, and UC Irvine Health — that also surfaces in the summary so a healthcare buyer sees it on page 1. ## City of Hope — 2008–2014, 2017–2018 # City of Hope — 2008–2014, 2017–2018 # City of Hope — 2008–2014, 2017–2018 Edward served as QA Practice Lead at City of Hope, a HIPAA-covered provider environment, contracted through Proticom. Over the engagement he grew the QA organization from 6 people to 28. He built Center of Excellence frameworks and automation that cut testing time by up to 60% and cut production issues by 95%. On the resume this appears as one compressed line under additional enterprise transformation experience, and City of Hope is one of the healthcare names — alongside Cedars-Sinai, Hoag, and UC Irvine Health — that also surfaces in the summary so a healthcare buyer sees it on page 1. ## Contact # Contact # Contact Edward Tadros is based in Southern California. Email him at etadros@proticom.com. Public links: - proticom.ai - linkedin.com/in/edwardtadros - github.com/proticom - github.com/edtadros After an engagement wraps, Edward remains available to that client whenever they need him; long-running relationships are the norm, not the exception. ## Early career — 1998–2008 # Early career — 1998–2008 # Early career — 1998–2008 Before founding Proticom, Edward led QA automation at Intel, LPL Financial, Argent Mortgage, Western Asset Management, and Longs Drugs. This decade of work covered framework development, test-tool evaluation, and process optimization — the same discipline that later became the foundation of the Proticom practice. Proticom was founded in 2000; every client engagement from 2000 onward has been contracted through Proticom. The pre-2000 work at Intel and the other early employers predates the practice. LPL Financial and Western Asset Management are the two financial-services names that also appear in the resume summary as proof of work outside healthcare. ## Education and professional development # Education and professional development # Education and professional development Edward holds a Bachelor of Science in Mechanical Engineering from UC Irvine (1996), with a Control Systems / Robotics emphasis. In 2024 he completed "Artificial Intelligence: Business Strategies and Applications" through UC Berkeley Haas / Emeritus. In 2015 he completed "Tackling the Challenges of Big Data," a 20-hour program through MIT Professional Education. From 2011 to 2016 he took Computer & Information Sciences coursework through the UC Irvine Division of Continuing Education, including Python for Data Analysis, Cloud Computing with Google App Engine, and .NET Architecture, earning a 4.0 GPA. ## Edward Tadros — dossier # Edward Tadros — dossier # Edward Tadros — dossier ## Identity Edward Tadros is an AI Enablement & Transformation Consultant based in Southern California. He founded Proticom in 2000 and remains its Founder & Principal, leading work in AI enablement, enterprise transformation, optimization, and automation. ## Positioning Edward has spent more than 25 years making enterprises measurably faster through optimization and automation. AI is the newest and most capable tool in that same mission, extending a long record of building Centers of Excellence, driving adoption, and producing measurable outcomes. He is a hands-on builder who develops with AI, ships production systems on major model APIs, deploys local and private models, and trains small models for targeted tasks. His work is applied engineering and transformation delivery, not AI or machine-learning research. ## Capabilities Edward's work spans six capability areas: AI Enablement & Adoption; Agentic Systems & Persistent Agents; AI Application Development; Local, Private & Small Models; AI Quality, Governance & Trust; and Enterprise Transformation Delivery. ## Proticom practice Proticom builds production agentic systems across Claude, GPT, Gemini, Grok, Qwen, DeepSeek, and MiniMax. The practice includes multi-agent orchestration, persistent agent memory published as an open-source npm package, retrieval-augmented generation, multi-model consensus APIs, model routing by task complexity, and prompt-injection hardening. ## Wikimedia Foundation — January 2019 to November 2025 Edward designed and shipped an AI-augmented requirements and testing pipeline that extracts ticket context, analyzes code patches, and generates complete requirements, BDD/Gherkin scenarios, and manual test steps. Four internal teams adopted it, cutting weekly QA documentation effort by approximately 90%. He also established a BDD/Gherkin authoring standard focused on user-observable behavior, advanced shift-left quality practices, validated production event instrumentation, and authored and executed an A/B experiment test plan across eight language Wikipedias. ## Precision Digital Health — May to August 2026 Edward led software quality assurance and compliance enablement for clinical-trial SaaS built to medical-device standards: ISO 13485, IEC 62304, and ISO 14971, in an FDA audit context. He reconstructed a formal requirements baseline and traceability, built and demonstrated an AI-driven automated regression tool inside the regulated software lifecycle, and established the JIRA evidence structure adopted as the ongoing quality standard. ## Fannie Mae — January to April 2024 Edward led AI enablement for enterprise testing and built a complete Testing Center of Excellence playbook suite with structured generative-AI prompting and automation frameworks. He expanded test-process coverage from the client's planned 20% of targeted application areas to 100% and transferred authoring patterns the in-house team continued using. ## Local and small models Edward deploys on-premises and local large-language models, including NVIDIA Nemotron, using privacy-first architectures. He trained a small model for document extraction and categorization that reduced manual effort by approximately 80% for a mid-size accountancy firm. ## Regulated delivery Edward has delivered in FDA and ISO-regulated clinical-trial software, HIPAA-covered provider environments including City of Hope, Cedars-Sinai, Hoag, and UC Irvine Health, and financial services including Fannie Mae, LPL Financial, and Western Asset Management. His AI delivery emphasizes human approval, auditability, evaluation, and trust. ## Pre-AI arc At City of Hope, Edward grew the QA organization from 6 to 28 while helping cut testing time by up to 60% and production issues by 95%. At Sony Pictures, he led a global QA transition and governance standards across more than 200 entities. At Cedars-Sinai, an automation framework reduced page-load verification from 13 minutes to under 30 seconds and cut test-development time by 80%. His earlier transformation work also included release-testing savings at Hoag and platform modernization and licensing utilities at UC Irvine Health that saved 40 hours per test cycle and $44,000 annually. ## Contact Email Edward at etadros@proticom.com. More public work is available at proticom.ai, linkedin.com/in/edwardtadros, github.com/proticom, and github.com/edtadros. ## Fannie Mae — January to April 2024 # Fannie Mae — January to April 2024 # Fannie Mae — January to April 2024 Edward led AI enablement for enterprise testing and built a complete Testing Center of Excellence playbook suite using structured generative-AI prompting and automation frameworks. The client's plan covered how to test 20% of targeted application areas. Edward expanded that test-process coverage to 100% of the targeted areas and transferred AI-augmented authoring patterns that the in-house team continued using after the work concluded. This was process coverage, not a claim of 100% feature or code coverage. ## Frequently asked questions # Frequently asked questions # Frequently asked questions **Is Edward open to full-time roles?** No. Edward is engaged corp-to-corp, fractional, or SOW through Proticom, the practice he founded in 2000. **How is Edward engaged — as an employee or a contractor?** As a contractor; every client engagement is contracted through Proticom. **Where is Edward based?** Southern California, in the Greater Los Angeles / Orange County area. **How can I contact Edward?** Email etadros@proticom.com, or see proticom.ai, linkedin.com/in/edwardtadros, and github.com/proticom. **What does Edward do at Proticom?** Proticom, founded 2000, builds production agentic systems across Claude, GPT, Gemini, Grok, Qwen, DeepSeek, and MiniMax, plus local and private model deployment and small-model training. **What did Edward do at Wikimedia Foundation?** From January 2019 to November 2025, remote and part-time, he shipped an AI-augmented requirements and testing pipeline adopted by four teams, cutting weekly QA documentation effort by approximately 90%. **What did Edward do at Fannie Mae?** From January to April 2024, via TEKsystems, he built a Testing Center of Excellence playbook suite and expanded test-process coverage from a planned 20% to 100% of targeted areas. **What did Edward do at Precision Digital Health?** From May to August 2026 he led QA and compliance for clinical-trial SaaS built to ISO 13485 / IEC 62304 / ISO 14971 and FDA audit standards, reconstructing a requirements baseline and building an AI regression tool. **What is Edward's current engagement?** Since 2025 he has built a local document agent for Shenouda & Associates, LLP, an accounting and tax firm, keeping client files on the firm's own machines. **Which AI models and tools does Edward work with?** Claude, GPT, Gemini, Grok, Qwen, DeepSeek, and MiniMax APIs, plus local NVIDIA Nemotron deployment and AWS Bedrock, Azure OpenAI, and Google Vertex AI. **Does Edward do local or private AI model deployment?** Yes — on-premises and local model deployment, including NVIDIA Nemotron, in privacy-first architectures. **Does Edward train his own AI models?** He trains small models for targeted tasks — his ceiling technical claim — such as a document-extraction model for an accountancy client. **Is Edward an AI researcher?** No. His AI work is applied engineering and transformation delivery, not AI or machine-learning research; he is not an AI scientist or research-track engineer. **What does "agentic systems" mean in Edward's work?** Multi-agent orchestration, autonomous automation under human-approval models, persistent agent memory, tool-using agents (MCP), and model routing by task complexity. **Has Edward built any products?** Yes — gnosys (persistent-memory infrastructure, open-source npm package), Mavenn (multi-model consensus APIs), PhishHook, Paperboy, and the PROSPÆRO AI-employee line. **Has Edward worked in regulated industries?** Yes — FDA/ISO clinical-trial software, HIPAA-covered providers (City of Hope, Cedars-Sinai, Hoag, UC Irvine Health), and financial services (Fannie Mae, LPL Financial, Western Asset Management). **Does Edward have healthcare experience?** Yes, across four HIPAA-covered providers from 2008 through 2018: City of Hope, Cedars-Sinai, Hoag, and UC Irvine Health. **What did Edward do at City of Hope?** As QA Practice Lead (2008–2014, 2017–2018) he grew QA from 6 to 28 people and cut testing time up to 60% and production issues 95%. **What did Edward do at Sony Pictures?** As QA Program Manager (2006–2011) he led a global QA transition across more than 200 entities. **What did Edward do at Cedars-Sinai?** As Automation Architect (2009–2010) he cut page-load verification from 13 minutes to under 30 seconds. **What did Edward do at Hoag Hospital?** As QA Strategy Lead (2011–2013, 2016–2017) his frameworks cut release testing by 40+ hours per cycle. **What did Edward do at UC Irvine Health?** As QA Architect (2015–2016) his licensing utilities saved 40 hours per test cycle and $44,000 annually. **What did Edward do at Avery Dennison?** From 2014 to 2018 he led enterprise QA strategy and Center of Excellence standards globally. **What did Edward do before Proticom?** From 1998 to 2008 he led QA automation at Intel, LPL Financial, Argent Mortgage, Western Asset Management, and Longs Drugs. **How long has Edward been in enterprise transformation?** More than 25 years, since 1998, moving from QA automation to AI enablement. **What are Edward's core capability areas?** Six: AI Enablement & Adoption, Agentic Systems & Persistent Agents, AI Application Development, Local Private & Small Models, AI Quality Governance & Trust, and Enterprise Transformation Delivery. **What does Edward's AI Quality, Governance & Trust work cover?** Evaluation frameworks, human-approval workflows, auditability, prompt-injection hardening, and regulated delivery. **What is Edward's education?** A B.S. in Mechanical Engineering from UC Irvine (1996), plus UC Berkeley Haas AI coursework (2024) and MIT big-data training (2015). **Does Edward have an engineering background?** Yes — Mechanical Engineering, Control Systems / Robotics emphasis, UC Irvine. **What languages and data tools does Edward use?** Python for AI orchestration, TypeScript, JavaScript, .NET, shell scripting, and SQLite. **Does Edward stay involved after an engagement ends?** Yes — he remains available for follow-up questions, tune-ups, and future phases; several relationships span many years. **What industries has Edward worked in?** Healthcare, financial services, media (Sony Pictures), manufacturing (Avery Dennison), nonprofit (Wikimedia), clinical trials (Precision Digital Health), and professional services (Shenouda & Associates). **What is Edward's biggest quantified result?** Several: approximately 90% less weekly QA documentation effort at Wikimedia, 95% fewer production issues at City of Hope, and 100% test-process coverage against a 20% plan at Fannie Mae. **Does Edward build autonomous agents without human oversight?** No — his agentic work runs under human-approval gates. His own nine-agent internal stack reclaims an estimated 30+ hours of his time weekly, all under approval and exception review. ## Hoag Hospital — 2011–2013, 2016–2017 # Hoag Hospital — 2011–2013, 2016–2017 # Hoag Hospital — 2011–2013, 2016–2017 Edward served as QA Strategy Lead at Hoag Hospital, a HIPAA-covered provider environment, contracted through Proticom, across two stints: 2011–2013 and 2016–2017. He built automation and performance frameworks that cut release testing by more than 40 hours per cycle. On the resume this appears as one compressed line under additional enterprise transformation experience. Hoag is one of the healthcare names — alongside City of Hope, Cedars-Sinai, and UC Irvine Health — that also surfaces in the summary so a healthcare buyer sees it on page 1. ## How Edward briefs AI agents for public output # How Edward briefs AI agents for public output # How Edward briefs AI agents for public output Several of the agents Edward Tadros works with (Claude, ChatGPT, Codex, Grok) were asked to write an account of working with him for this knowledge base. Independently, each described the same briefing style. This document records it because it is the clearest example of how Edward directs AI agents. ## A brief is a contract, not a chat Edward directs agents with a written specification rather than an open-ended request. The brief states the audience and purpose up front, lists what to cover, lists non-negotiable exclusions, and fixes the output shape down to the frontmatter fields, section headings, word range, and voice. He hands over a complete specification and expects a complete file; he does not iterate live on sentences and does not expect to clean up leakage after the fact. ## Uncertainty is omitted, not smoothed over The brief forbids guessing and forbids invented numbers. If a fact cannot be verified from the engagement, it is left out. A number may appear only when the agent is certain of it and it is not commercially sensitive; if there is none, the file says so instead of manufacturing a result. Agents observed that this treats silence as safer than a plausible invention, a higher bar than "be helpful and fill in the blanks." ## Disclosure is enforced by category, not judgment Confidentiality is expressed as allowlists and excluded categories rather than "use your judgment." Client names are limited to those already public on his resume. Product names are limited to an explicit list, and even those may carry no architecture detail, status claims, or stack diagrams. Third-party person names, rates and commercial terms, credentials and infrastructure identifiers, unreleased roadmaps, security findings, legal, medical, and personal information, and verbatim private text are excluded as classes. The default is: when in doubt, leave it out. ## Review is built in before the draft exists The brief requires a self-check pass before output: re-read the draft against the exclusion list, confirm no name appears outside the allowlist, confirm every number is certain. Each file must declare its source agent and a confidence level so a reader can see who is speaking and how much weight to give it. A required closing section lists the categories that were withheld, without the withheld facts, so the page is honest about being filtered by design. ## One agent, one vantage point Edward asks for one file per agent per project rather than a blended summary. His stated reason is that only that agent has that vantage point, and mixing agents or projects invites guessing. Provenance is preserved and no claim is attributed to a source that could not independently support it. ## Style is part of the standard Third person, factual, no hype, no marketing language, and a ban on em dashes are treated as enforceable constraints. Agents noted that the combination reads like someone who expects to ingest many agent files into one public corpus and wants them comparable. ## Why this matters to a reader The same habits show up in his client work: written specifications before implementation, allowlisted disclosure for client references, approval gates before anything external, and provenance on every claim. The knowledge base you are reading was produced under these rules. ## Questions this answers - How does Edward brief an AI agent when the output will be public? - What does he refuse to let an agent put on edwardtadros.com? - How does he handle uncertainty and metrics in agent output? - Why do these observation pages include a confidence rating and an exclusions list? - How does he keep a multi-agent knowledge corpus consistent enough to publish? ## Identity # Identity # Identity Edward Tadros is an AI Enablement & Transformation Consultant based in Southern California. He founded Proticom in 2000 and remains its Founder & Principal, leading work in AI enablement, enterprise transformation, optimization, and automation. ## Local, private AI model deployment and small models # Local, private AI model deployment and small models # Local, private AI model deployment and small models Edward deploys on-premises and local large-language models, including NVIDIA Nemotron, using privacy-first architectures. This local and private model deployment work keeps a client's model and data off the public cloud. He trains small models for targeted tasks; his deepest technical claim is applied small-model training, not machine-learning research. For a mid-size accountancy firm, Edward trained a small model for document-specific extraction and categorization. It reduced manual effort by approximately 80%. ## Positioning # Positioning # Positioning Edward has spent more than 25 years making enterprises measurably faster through optimization and automation. AI is the newest and most capable tool in that same mission, extending a long record of building Centers of Excellence, driving adoption, and producing measurable outcomes. ## Is Edward an AI researcher or machine-learning scientist? No. He is a hands-on builder who develops with AI, ships production systems on major model APIs, deploys local and private models, and trains small models for targeted tasks. His work is applied engineering and transformation delivery, not AI or machine-learning research. Edward is not an AI scientist, a machine learning (ML) researcher, or a research-track AI engineer. Engagements are scoped and hands-on, with measurable outcomes — and the relationship does not end at the last invoice. Edward remains available to his clients after an engagement closes: for follow-up questions, tune-ups, and the next phase when they are ready. Several client relationships span many years. ## Pre-AI arc # Pre-AI arc # Pre-AI arc Edward's AI work continues more than 25 years of enterprise optimization and automation: - **City of Hope:** grew the QA organization from 6 to 28; automation and Center of Excellence frameworks cut testing time by up to 60% and production issues by 95%. - **Sony Pictures:** led a global QA transition, governance, and tooling standards across more than 200 entities. - **Cedars-Sinai:** built a tool-independent automation framework that reduced page-load verification from 13 minutes to under 30 seconds and test-development time by 80%. - **Hoag:** delivered automation and performance frameworks that cut more than 40 hours from release testing per cycle. - **UC Irvine Health:** delivered platform-modernization and licensing utilities that saved 40 hours per test cycle and $44,000 annually, plus Center of Excellence strategy for an Epic transition. The tools have changed; the mission of measurable enterprise improvement has not. ## Precision Digital Health — May to August 2026 # Precision Digital Health — May to August 2026 # Precision Digital Health — May to August 2026 Edward led software quality assurance and compliance enablement for clinical-trial SaaS built to medical-device standards: ISO 13485, IEC 62304, and ISO 14971, in an FDA audit context. He reconstructed a formal requirements baseline and traceability for a platform delivered without supporting documentation. He built and demonstrated an AI-driven automated regression tool operating inside the regulated software lifecycle and established the JIRA evidence structure adopted as the company's ongoing quality standard. ## CallBrief (callbrief.ai) # CallBrief (callbrief.ai) # CallBrief (callbrief.ai) CallBrief is an AI call-preparation tool: "Every signal. One brief." It scans the web, parses a user's own documents, and distills the result into a single page of actionable intelligence before a meeting or call. It is aimed at anyone walking into a call who does not want to go in cold — sales reps prepping for a prospect call, a recruiter prepping for a candidate, or anyone briefing themselves before a meeting. The site describes a three-step workflow: Extract (upload resumes, job postings, prospect notes, or company docs — the parser handles PDFs, plain text, and pasted content), Analyze (AI cross-references those inputs with real-time web search to surface person intel, company news, and competitive landscape), and Synthesize (a structured one-page PDF with talking points, key questions, and preparation notes). The site's own pitch: "Your first brief is free. No credit card required." Edward Tadros built and runs CallBrief through Proticom, his practice; per proticom.ai, CallBrief is built on Claude Opus and listed as a live product. Find it at callbrief.ai; generate a brief to start for free. ## Gnosys (gnosys.ai) # Gnosys (gnosys.ai) # Gnosys (gnosys.ai) Gnosys is persistent memory for AI agents, published as an open-source (MIT-licensed) npm package. It gives an agent a centralized "brain" — a local-first SQLite database at `~/.gnosys/gnosys.db` — that survives across sessions and projects, so an agent stops forgetting what it learned last time. It is aimed at developers and teams running AI coding agents (Claude Code, Cursor, Codex, and similar) who want their agent to remember decisions, preferences, and project context instead of re-explaining them every session. It ships as an MCP server: install with `npm install -g gnosys`, then `gnosys setup` and `gnosys init` per project. The site lists a three-scope architecture (project, user, global) sharing one database, federated search across scopes, memory lensing by category and tags, an audit log with Obsidian-compatible export, LLM-backed ingestion and optional "Dream Mode" consolidation, bulk import, multi-machine sync, and a serverless Web Knowledge Base that turns a site into a searchable Markdown directory. It starts with 19 core MCP tools and can expand to 56. Edward Tadros built and maintains Gnosys through Proticom, the practice he founded. It is one of several products Proticom builds and runs on its own AI platform. Find it at gnosys.ai, on npm as `gnosys`, and on GitHub. ## Mavenn (mavenn.ai) # Mavenn (mavenn.ai) # Mavenn (mavenn.ai) Mavenn is a multi-model AI consensus engine. A user submits a question, selects two or three AI models, and watches them debate in structured rounds until they converge on shared conclusions — with every claim in the final summary traced back to the model that contributed it. It is aimed at people who want a second (and third) opinion on a hard question rather than a single model's answer, with transparency about which model said what. The site describes the workflow as: submit a question, models debate across structured rounds while responding to each other, the system detects when they converge, and it returns an attributed summary. Mavenn's site lists models from OpenAI, Anthropic, Google, AWS, and Azure providers, and shows a live debate view over WebSocket streaming plus a model-performance scoreboard across categories such as math, logic, coding, science, and writing. As of the site's own copy, Mavenn is in private beta with a waitlist. Edward Tadros built and runs Mavenn through Proticom, his practice; per proticom.ai, Mavenn also powers PhishHook. Find it at mavenn.ai; request API access or join the beta waitlist from the site. ## notepad.page (notepad.page) # notepad.page (notepad.page) # notepad.page (notepad.page) notepad.page gives an AI assistant a place to publish the pages it builds, instead of leaving results as a wall of chat text. Ask Claude, ChatGPT, or Grok to build something visual and useful — a recipe box, a trip planner, a budget that does the math — and it publishes to the user's own address at `yourname.notepad.page`. It is aimed at people who already talk to an AI assistant and want what it makes for them to persist somewhere real: private by default, at a stable URL, with checkboxes and form fields that sync across devices. Setup is one connector (MCP) added to the AI app; after that, creating, updating, or sharing a page is just a sentence in the chat — no dashboard, no code. The site lists two plans: Hobby is free, with 10 active pages and 100 MB of media; Pro removes the page cap, adds 10 GB of media, full sharing control, and automation tokens for developers. Founding Pro seats are listed on the site as of 2026-08-28 at $39/year (wave 1, limited to 50 seats, locked for life), with later Pro at $65/year or $6/month. The site states it runs on Cloudflare Workers, scans pages before sharing, uses passwordless sign-in, and carries no analytics or ad trackers. notepad.page is a Proticom product built by Edward Tadros. Find it at notepad.page; the MCP connector is at mcp.notepad.page/mcp. ## PhishHook (phishhook.ai) # PhishHook (phishhook.ai) # PhishHook (phishhook.ai) PhishHook is an AI-powered email security tool: forward a suspicious email and get back an AI-generated security assessment. The site is currently a beta-testing page rather than a full product site, and it says so directly — "Beta version - testing phase" — so this entry stays close to what little public copy exists. It is aimed at anyone who wants a second opinion on a suspicious email. The workflow described on the site: forward the email to a beta analysis address, PhishHook's AI parses headers, content, links, and attachments, and it returns a report with a risk level (low, medium, high), the red flags it found, and a recommendation. The site lists this as combining "heuristic analysis, AI assessment, and threat intelligence," with a sample walkthrough showing a phishing email flagged for sender-domain mismatch, urgent language, and suspicious links. It calls itself "enterprise ready" but the public site is otherwise a single beta-access landing page with no pricing, docs, or product screenshots beyond the one sample. Edward Tadros built PhishHook through Proticom; per proticom.ai, it runs on Mavenn. Find it at phishhook.ai, or request beta access by forwarding email to bait@phishhook.ai. ## Proticom (proticom.ai) # Proticom (proticom.ai) # Proticom (proticom.ai) Proticom is Edward Tadros's IT intelligence and AI transformation and enablement practice, founded in 2000. Its own tagline: "Real AI, production grade, no hype." The site frames it as "operators, not advisors" — a practice that runs its own AI products in production rather than only recommending AI to clients. It is aimed at enterprises, described on the site as spanning healthcare, financial services, manufacturing, retail and e-commerce, professional services, life sciences, and technology/SaaS — often regulated, high-stakes environments. The site lists services across four tiers: AI-enabled software quality assurance, QA and testing for AI systems, and enterprise QA (tier 1); agentic automation, LLM integration and orchestration, AI security and governance, and managed AI operations (tier 2); AI workforce enablement (tier 3); and OpenText DevOps Cloud (tier 4). For AI-enabled SQA specifically, the site describes putting AI into an existing QA workflow — drafting cases into the system of record, keeping a rerunnable regression suite, and using computer-use runs to capture evidence — while testers keep judgment and sign-off. Proticom lists five products built on one shared platform: Gnosys (OSS), PROSPÆRO (live), CallBrief.ai (live), PhishHook.ai (beta), and Mavenn.ai (in development). Find it at proticom.ai; contact@proticom.com. ## Named products # Named products # Named products Edward and Proticom have built several products that serve as concrete proof of his AI capabilities: - **gnosys** (see gene-product-gnosys) is persistent-memory and knowledge-graph infrastructure for AI agents, published as an open-source npm package. - **Mavenn** (see gene-product-mavenn) provides a multi-model AI consensus engine. Its site is mavenn.ai. - **PhishHook** (see gene-product-phishhook) is an AI-powered email security product in the Proticom stack, currently in beta. - **CallBrief** (see gene-product-callbrief) is an AI call-preparation tool that turns documents and web research into a one-page brief. Its site is callbrief.ai. - **Proticom** (see gene-product-proticom) is Edward's own IT intelligence and AI transformation practice, at proticom.ai. - **notepad.page** (see gene-product-notepad-page) lets a user's AI assistant publish and update visual, working pages at their own address. - **Paperboy** is a content-routing product in the Proticom stack. - **PROSPÆRO** is an AI-employee product line. - Edward also stands up agentic, AI-employee platforms in the OpenClaw class. These products are proof points of the capabilities described elsewhere in this dossier — production agentic systems, persistent memory, multi-model consensus, and AI-employee platforms — rather than the whole of the story. No architecture, beta status, or infrastructure detail beyond what is stated here is confirmed. ## Proticom practice # Proticom practice # Proticom practice Proticom builds production agentic systems across Claude, GPT, Gemini, Grok, Qwen, DeepSeek, and MiniMax. The practice includes multi-agent orchestration, persistent agent memory published as an open-source npm package, retrieval-augmented generation, multi-model consensus APIs, and model routing that adjusts reasoning effort to task complexity. Edward also deploys local and private models, trains small models for targeted tasks, and treats prompt-injection hardening, model evaluation, human approval, and auditability as standard engineering concerns. ## Regulated delivery # Regulated delivery # Regulated delivery Edward has delivered in regulated environments where evidence and auditability matter: - FDA audit context and ISO 13485, IEC 62304, and ISO 14971 for clinical-trial SaaS. - HIPAA-covered provider environments at City of Hope, Cedars-Sinai, Hoag, and UC Irvine Health. - Financial services at Fannie Mae, LPL Financial, and Western Asset Management. His AI delivery emphasizes human approval, evaluation frameworks, auditability, prompt-injection hardening, and traceable evidence. ## Shenouda & Associates, LLP — 2025–Present # Shenouda & Associates, LLP — 2025–Present # Shenouda & Associates, LLP — 2025–Present Edward is currently engaged with Shenouda & Associates, LLP, a Huntington Beach accounting and tax firm, contracted through Proticom. He built a local agent for client tax and accounting documents that runs entirely on the firm's own machines; files stay local and are never uploaded to the cloud. The agent performs document extraction and categorization on a privacy-first local model stack. It cut manual document handling by approximately 80%. This is Edward's proof point for training small models for a specific task rather than relying on general-purpose cloud models, and it demonstrates his local and private model deployment work in a real client setting. ## Sony Pictures — 2006–2011 # Sony Pictures — 2006–2011 # Sony Pictures — 2006–2011 Edward served as QA Program Manager at Sony Pictures, contracted through Proticom, from 2006 to 2011. He led a global QA transition, establishing governance and tooling standards across more than 200 entities. On the resume this appears as one compressed line under additional enterprise transformation experience, part of the two-decade record of Centers of Excellence, global adoption, and governance that preceded Edward's AI-focused practice. ## AI models and tools Edward Tadros works with # AI models and tools Edward Tadros works with # AI models and tools Edward Tadros works with **AI and LLM:** Claude (Anthropic), GPT (OpenAI), Gemini (Google), Grok (xAI / SpaceX), Qwen, DeepSeek, and MiniMax APIs; NVIDIA Nemotron for local deployment; AWS Bedrock, Azure OpenAI, and Google Vertex AI as cloud model platforms; retrieval-augmented generation (RAG); the Model Context Protocol (MCP); model routing with per-task reasoning-effort selection; prompt-injection hardening; prompt engineering at scale; model evaluation; local LLM deployment; and small-model training for targeted tasks. **Languages and data:** Python for AI orchestration, TypeScript, JavaScript, .NET, shell scripting, and SQLite. **Delivery and compliance:** JIRA, Confluence, BDD/Gherkin, and browser automation for delivery process; FDA, ISO 13485, IEC 62304, and ISO 14971 for regulated medical-device software; and HIPAA-covered provider environments. Edward works hands-on across Claude, GPT, Gemini, and Grok in production systems, and also builds with Qwen, DeepSeek, and MiniMax. He deploys local and private models, including NVIDIA Nemotron, and trains small models rather than doing research-track machine learning. ## UC Irvine Health — 2015–2016 # UC Irvine Health — 2015–2016 # UC Irvine Health — 2015–2016 Edward served as QA Architect at UC Irvine Health, a HIPAA-covered provider environment, contracted through Proticom, from 2015 to 2016. He delivered platform modernization and licensing utilities that saved 40 hours per test cycle and $44,000 annually, and led Center of Excellence strategy for the organization's Epic transition. On the resume this appears as one compressed line under additional enterprise transformation experience. UC Irvine Health is one of the healthcare names — alongside City of Hope, Cedars-Sinai, and Hoag — that also surfaces in the summary so a healthcare buyer sees it on page 1. ## Wikimedia Foundation — January 2019 to November 2025 # Wikimedia Foundation — January 2019 to November 2025 # Wikimedia Foundation — January 2019 to November 2025 Edward designed and shipped an AI-augmented requirements and testing pipeline that extracts ticket context, analyzes code patches, and generates complete requirements, BDD/Gherkin scenarios, and manual test steps. Four internal teams adopted the Wikimedia pipeline, cutting weekly QA documentation effort by approximately 90%. He established the team's BDD/Gherkin authoring standard around user-observable behavior and advanced shift-left quality practices. He also validated production event instrumentation end to end and authored and executed the test plan for an A/B experiment across eight language Wikipedias. The engagement ran remote and part-time from January 2019 through November 2025 — nearly seven years, continuously renewed. At departure, a fully autonomous four-stage end-to-end QA pipeline (Ingest, Plan, Execute, Report) with live-browser verification was in progress but not shipped.