A Forward Deployed Engineer who already ships frontier AI into production.
Forward Deployed Engineer is the work Fauzul has been doing for a decade — dropping into ambiguous enterprise environments (health-tech, fintech, HR-tech, hospitality, export) and pulling production software out of fuzzy briefs. Over the last year that became shipping agents: a Claude-powered natural-language CMS running in production for a 26-year-old export house, plus agentic 0→1 builds (, , ) using multi-agent reasoning (A2A, MCP, ReAct) and n8n orchestration. He owns the full scope — discovery, scoping, system design, build, and production rollout, flexing across frontend and backend — and treats reliability, compliance and auditability (HIPAA, GDPR, OIDC/RBAC) as product surfaces. He's intentional about being on-site in Dublin, embedded with customers — exactly the profile OpenAI's Forward Deployed Engineering team needs to take frontier models into real-world enterprise production.
Frontier AI, already in production
Forward Deployed Engineer is the work I'd already been doing. For a decade I've dropped into ambiguous enterprise environments and pulled production software out of fuzzy briefs — and over the last year that became shipping agents into real client stacks.
An LLM-powered CMS, running in a client's production stack
For — a 26-year-old Dhaka export house — I skipped the traditional headless CMS entirely. Instead I deployed an LLM-based system: the client expresses the change they want in plain natural language, the model turns that into a pull request, and CI/CD ships it to production at the edge. Natural-language intent in, reviewed and reversible change out — frontier AI doing real work in a legacy, specification-sensitive industry. The deployment pattern is model-agnostic; the FDE skill is what makes it land.
Natural-language intent → model-authored PR → automated CI/CD deploy to production at the edge.
Every change stays inside Git — reviewable, auditable, revertible. No developer middleman, no CMS lock-in.
Agentic 0→1 in public: (goal-first intros), (autonomous briefings via n8n + LLM steps), — multi-agent reasoning with A2A and MCP.
Reliability and compliance treated as product surfaces — HIPAA, GDPR/VAT, OIDC/RBAC — so models can touch sensitive workflows with isolation and auditability built in.
LLM-powered CMS — in production
Natural-language requests become model-authored PRs and automated deploys for a 26-year-old export house. Real agentic AI in a legacy industry.
Agentic 0→1 — TagRamp · NewScriber · VisaPros
Multi-step agents that plan and execute (ReAct-style), orchestrated with A2A, MCP and n8n — real work, not demos.
Compliance by design — EKAGRA · Anygum
HIPAA-compliant EHR pipeline and federated OIDC + instance-level RBAC. Auditable and isolated from day one.
Point by point
The role is Forward Deployed Engineer — leading complex, end-to-end deployments of frontier models in production alongside OpenAI's most strategic customers: discovery, scoping, system design, build, rollout, and eval-driven feedback that shapes product and model roadmaps. Here is how a decade of end-to-end enterprise delivery and a year of shipping agents into real client stacks maps onto each point. Pick a topic to explore the evidence.
Own technical delivery across multiple deployments — from first prototype to stable production — driving production adoption and measurable workflow impact.
Ten years founding and leading , owning use cases end-to-end from empty canvas → CI/CD pipeline → production across London, Zurich, Oslo, Sandnes, Dubai, Florida and Quebec. Ships 0→1 across six verticals and multiple continents — health-tech, fintech, HR-tech, hospitality, export, and now esports infrastructure (, unifying Bangladesh's fragmented Dota 2 scene into one national hub) — and measures success by what reaches production and sticks: validating demand before over-building (ran four pop-up events to test a restaurant brand before committing capital), then framing the work around the outcome.
Build full-stack systems that deliver customer value, contributing directly in the code — across frontend and backend — when progress or clarity depends on it.
End-to-end builder who still writes the code. Started as a UX Engineer and ships polished frontends (React, Next.js, Angular, Tailwind) alongside Go/Node/TypeScript backends, so flexing into whatever the deployment demands — frontend, backend, infra or glue — is native, not a stretch. A Claude-powered natural-language CMS for turns plain-English intent into reviewed PRs and automated edge deploys.
Embed closely with customer teams — understand their needs, scope and sequence delivery, remove blockers early, and guide adoption of what you build.
A decade dropping into ambiguous corporate environments and translating raw requirements into granular engineering tickets, leading delivery by example and earning trust with both engineers and executives. Comfortable in the room with the customer — that on-site, embedded mode is the part of the job he wants, not a cost he tolerates.
5+ years of engineering or technical deployment experience that includes customer-facing work, scoping and delivering complex systems in fast-moving or ambiguous environments.
Co-founded and ran it for a decade — grew the team to 22, then to a distributed remote-first firm delivering for clients in London, Zurich, Oslo, Dubai, Florida, Montreal and Laval. Cut time-to-market dramatically under real constraints and led products into GDPR, HIPAA and WCAG 2 compliance across five client organisations — the definition of customer-facing delivery in fast-moving, ambiguous environments.
Write and review production-grade code across frontend and backend using Python, JavaScript, or comparable stacks — clean, testable, observable, scalable.
JavaScript/TypeScript and Go are Fauzul's production core, with Python in his AI and data work — squarely inside the 'Python, JavaScript, or comparable stacks' the role asks for. He already ships the way the JD describes: typed contracts, CI gates, Jest/Cypress/Playwright suites, and structured logging, so clean, testable, observable, scalable code is the default rather than the aspiration.
Have built or deployed systems powered by LLMs or generative models, and understand how model behaviour affects product experience.
Ships agentic, multi-step systems in public and in production — autonomously researches, plans and distributes briefings via orchestrated n8n + LLM steps; plans and executes goal-first networking intros; uses multi-agent reasoning (A2A, MCP). He designs around model behaviour — prompt and context design, tool-calling, retrieval, and human-in-the-loop on high-stakes steps — because the model's quirks are the product's edges.
Drive eval-driven feedback — measuring workflow impact and model quality so the field sharpens what OpenAI's Research and Product teams build next.
Builds technical evals and POCs to qualify what ships — for ' multi-agent visa advisor he engineered mitigation frameworks like semantic-chunking hybrid retrieval and asynchronous evaluation guardrails to hold up reliability in production — and brings a strong automated-testing and compliance-auditability discipline (unit, integration, E2E). What's still missing is a formal, metric-driven LLM-eval harness as a first-class, roadmap-shaping product surface rather than per-project guardrails.
His instinct already points here: treat model quality the way he treats compliance — designed in, measurable, auditable from day one. He'd formalise golden eval sets, captured traces, and accuracy/safety/latency scorecards wired into CI, then close the loop back to Research and Product as structured field feedback — a 4–8 week ramp on habits he already has.
Strengths & honest gaps
Frontier AI already shipping in production
Replaced a traditional CMS with an LLM-powered natural-language system for — the client edits the live site in plain English, the model opens a PR, and CI/CD ships it to the edge. Alongside , and , that's real agentic AI doing real work in a legacy industry today, not a demo. The deployment skill is model-agnostic — exactly what an FDE brings.
A true Forward Deployed Engineer — owns the full scope
Sees a use case from ambiguous brief to CI pipeline to production, flexing into frontend, backend, infra or GTM as the problem demands. A decade dropping into enterprise environments across London, Zurich, Oslo, Dubai, Florida and Quebec, shipping 0→1 every time — the end-to-end ownership the FDE role is built around.
Customer-facing in regulated, high-stakes environments
Treats reliability, compliance and auditability as product surfaces, not back-office tax: a HIPAA-compliant EHR pipeline at (2,212 new patients onboarded, 9,650 schedules managed in 2026), GDPR + EU VAT commerce at — built from a blank canvas and grown over six years until Loopcloud acquired it in 2026 — and federated OIDC + instance-level RBAC at . The exit is the proof point: strategic customers bet on software built with this level of trust and predictability, and it paid off.
Calm judgment, honestly applied
Curious enough to live on the frontier (agentic builds in hackathons, learning in public, a 931-day French streak); cautious enough to probe risk early, bake in safeguards and tests, and be honest about gaps rather than oversell them. That's the temperament the role names — simplify complexity, decide fast under pressure, and model calm when the stakes are high.
Formal, metric-driven LLM-evaluation frameworks as a first-class, roadmap-shaping surface.
4–8 weeksHe already builds technical evals/POCs and holds a strong automated-testing and compliance-auditability discipline. He'd extend it into golden eval sets, captured traces and accuracy/safety/latency scorecards wired into CI — then feed that back to Research and Product as structured field signal, the same 'designed-in, not bolted-on' instinct he brings to compliance.
What else I bring
Strategic partner, developer champion, and technical operator. Fauzul’s background as a co-founder and CIO unlocks value far beyond the standard engineering parameters.
Developer Advocacy & Platform Champion
Fauzul has a proven track record of writing in public (blogging, tutorials) and maintaining a 900+ day learning streak. He is ready to act as a developer champion inside strategic customer teams, build high-fidelity demos/guides that highlight OpenAI's API capabilities, and help customer devs ship agents with confidence.
Founder-Operator Strategy Partnership
Having served as co-founder and CIO of multiple ventures, Fauzul operates with strong commercial and product empathy. He knows how to translate developer and customer friction directly into prioritized product requirements and feed those insights straight back to OpenAI's Research and Product teams to shape model and API roadmaps.
Compliance & Enterprise Gating
With deep compliance delivery experience (HIPAA pipelines at EKAGRA, GDPR/VAT structures at Jamahook), Fauzul speaks the language of corporate risk, privacy, and architecture reviews. He can proactively help customer security officers audit and approve OpenAI's trust boundaries, accelerating the path to production.
Candid answers
Candid answers — why a founder-operator wants the FDE seat, the honest read on shipping Claude and Gemini rather than GPT in production, how he builds eval rigor for agent quality, and how he's set up for hybrid Dublin plus heavy customer travel. Ask the agent anything else.
Your production AI work has been Claude- and Gemini-led, not GPT. Why are you the right Forward Deployed Engineer for OpenAI?
I'll be honest rather than name-drop: what I've shipped in production has been Claude- and Gemini-led — , for instance, is a multi-agent visa advisor built on Google's ADK Agent-to-Agent protocol — and that integrity matters more to me than pretending otherwise. But the FDE job isn't about which model — it's about taking frontier models into the messy reality of enterprise: discovery, scoping, system design, rollout, and the eval-driven feedback loop back to Research and Product. That skill is model-agnostic, and I've proven it across models in production — an LLM-powered natural-language CMS that turns plain English into reviewed PRs and automated deploys, plus agentic 0→1 builds spanning A2A, MCP and n8n orchestration. I understand how model behaviour shapes product experience because I've designed around it across more than one model family. I'm genuinely eager to go deep on OpenAI's models and tooling, and I'd rather be the engineer who's candid about where he's coming from and ramps visibly than the one who oversells.
You co-founded and run your own product engineering venture firm — why step into a Forward Deployed Engineer seat at OpenAI?
Because FDE is the work I love most, stripped of the agency overhead. As CIO my role drifted into running a client-services business; the part that lights me up is being in the room with a customer, turning an ambiguous problem into a shipped system. OpenAI lets me do exactly that with frontier models behind me — and deploying those models into real enterprise production is where my decade of scar tissue is most leveraged. I'd rather help define that frontier from the inside than keep grinding it from the outside.
You haven't worked at a frontier AI lab. Why should we trust you to deploy our models for strategic customers?
Because this is a deployment role, not a research role — the job is carrying the lab's models into the messy reality of enterprise, and that's what I've done for ten years. I've already put an LLM into a client's production stack and shipped agentic products in public. The FDE team doesn't need another model trainer; it needs someone with deployment callouses who can earn a strategic customer's trust, scope a use case end-to-end, contribute in the code, and ship it reliably — then feed what the field learns back to Research and Product. That's the exact shape of my last decade.
The role expects production-grade code in Python, JavaScript, or comparable stacks. Where does your stack actually sit?
Squarely inside it. JavaScript/TypeScript and Go are my production core, and the JD names JavaScript and comparable stacks right alongside Python — so I clear the bar on day one, not after a ramp. Python is where I already do AI and data work, and the discipline the role actually asks for — clean, testable, observable, scalable — is language-agnostic and exactly how I ship: typed contracts, CI gates, Jest/Cypress/Playwright suites, structured logging. Where a deployment is Python-first, I carry those habits straight across; where it's TS or Go, I'm already at full speed.
How do you build evaluation that goes beyond trial and error to measure model and workflow quality?
The same way I treat compliance: designed in and measurable from day one, not bolted on. On , my multi-agent visa advisor, I already built per-project guardrails — semantic-chunking hybrid retrieval plus asynchronous evaluation checks to catch bad completions before they reached a user. What I'd formalise at OpenAI is turning that instinct into a first-class harness: pin down success criteria with the customer, build golden eval sets and captured traces so every change is graded against accuracy, safety and latency rather than vibes, wire those scorecards into CI so regressions are caught automatically, and separate offline evals from production telemetry to see real-world drift. I've run rigorous automated test suites and compliance audits for years; making model evals a roadmap-shaping surface — and feeding the signal back to Research and Product — is the natural extension, and it's where I'd invest early.
This role is hybrid in Dublin (3 days/week) with travel up to 50%. Are you set up for that?
Yes — I'm intentional about basing this chapter in Dublin, and being on-site with customers is the part of the job I want, not a cost I tolerate. I've delivered for London, Zurich, Oslo and Sandnes clients for years and am comfortable in the room with both engineers and executives. I relocate as soon as the visa processes (Ireland's Critical Skills Employment Permit), commit to local hours and the in-office cadence, and treat heavy customer travel as core to deploying well rather than a burden. Dublin also keeps me close to the European customers and the family and friends I'd be moving toward.
The stack & the builds
The stack and the builds that prove it — agentic pipelines, RAG, federated identity, and an LLM-powered system already running in a client's production stack. Explore any topic below.
Ready to deploy frontier AI?
Intentional about relocating to Dublin, Ireland — ready for the hybrid office cadence and on-site customer travel from day one.