An innovator-builder who codes, ships, and modernizes — right inside the customer's stack.
This is the job a founder-CIO already does, written as a spec: an innovator-builder embedded inside the customer who doesn't just advise but codes, debugs and jointly ships bespoke agentic solutions to production. Fauzul has lived exactly this for a decade — co-founding , growing it into a 22-person team across engineering, product and design, and rising UX Engineer → Principal Staff SWE → Engineering Manager → CIO while staying hands-on enough to set the technical bar himself. He carries frontier AI into legacy reality: a Claude-powered natural-language CMS running in a 26-year-old export house's production stack, and , a six-agent A2A mesh on Gemini 2.5 Flash Lite with Google Search grounding — Python + FastAPI end to end — built as his Google × Kaggle 5-Day AI Agents Intensive capstone. He's fluent in the exact stack the role names: multi-agent patterns (ReAct, hierarchical delegation), RAG and vector retrieval, MCP and tool-calling, evals and observability — and treats the production blockers this role exists to clear (integration complexity, data readiness, state management) as product surfaces, with HIPAA, GDPR and OIDC/RBAC shipped across five client organisations. He's been on Google's stack for years — Flutter early (, 2019), Firebase in production, and Cloud Run running his latest initiatives today. A builder with genuine business acumen — accountable for revenue and ROI as a founder — who has served everyone from founder-led startups to established enterprises, and is intentional about Dublin.
Bespoke agentic solutions, shipped inside the customer — including the legacy ones
An embedded builder who codes, debugs and jointly ships — that's been my mandate for a decade. I carry frontier AI into legacy reality, and I still write the code.
A six-agent Gemini mesh — and a Claude CMS modernizing a 26-year-old export house
Two proofs, one mandate. On the Google stack: , the Google × Kaggle 5-Day AI Agents Intensive capstone — six parallel country agents as FastAPI microservices on Gemini 2.5 Flash Lite, fed by an accuracy-critical sequential parser and synthesised by a strategic advisor (hierarchical delegation), grounded against live Google Search at evaluation time. In legacy-industry production: a Claude-powered natural-language CMS for a 26-year-old export house, where plain English becomes model-authored pull requests and automated edge deploys — frontier AI running in a real customer's day-to-day operations, reviewable and revertible. The connective tissue between AI products and a customer's live infrastructure, built and shipped.
: A2A hub-and-spoke multi-agent architecture (ReAct + hierarchical delegation) on Gemini 2.5 Flash Lite with Google Search grounding — Python + FastAPI, MIT-licensed, the Google × Kaggle capstone.
Claude-powered natural-language CMS modernizing — plain English → model-authored PRs → CI/CD edge deploys, every change kept in Git.
Production blockers cleared: integration complexity, data readiness and state management — HIPAA pipeline at (2,212 patients · 9,650 schedules), Vipps/Stripe integrations, OIDC/RBAC multi-tenancy.
A decade on Google's stack: Flutter early (, 10k users, 2019) · Firebase in production · Cloud Run today · Gemini, ADK, Antigravity, on-device Gemma.
Innovator-builder — agentic systems in production
Codes, debugs and ships bespoke agentic solutions: VisaPros (A2A hub-and-spoke, ReAct, hierarchical delegation) on Gemini 2.5 Flash Lite + Google Search grounding · Claude-powered natural-language CMS in a client's stack · MCP & tool-calling · RAG, embeddings & vector search (pgvector) · evals & observability he builds himself.
Legacy modernization & business acumen — startups to enterprise
Carries frontier AI into legacy reality (Global Jute, 26-yr export house) · founder accountable for revenue & ROI · discovery and post-sales delivery across five verticals from founder-led startups to established enterprises · clears integration, data-readiness and state-management blockers (HIPAA, Vipps/Stripe, OIDC/RBAC).
A decade on Google's stack — early and ongoing
Flutter early (Singistic, 10k users, 2019) · Firebase in production (notifications, test distribution, a Firestore migration with A/B-tested onboarding for a Dubai client) · Cloud Run running today's initiatives · Gemini · Google ADK · Antigravity daily · on-device Gemma 4 E2B in this site's chat · Google × Kaggle Agents Intensive.
Qualification by qualification
Mapped point-by-point to the posting's minimum and preferred qualifications — verbatim, with a verdict and evidence for each. Where there's a real gap, it's marked as one, with the bridge.
Minimum qualifications
Bachelor's degree in Engineering, Computer Science, a related field, or equivalent practical experience.
Field of study: Computer Science & Engineering at North South University — then the equivalent-practical-experience route the posting explicitly names: a UN internship (2016–17), co-founding in 2017, and nine-plus years of shipped production systems since.
5 years of experience with software development using Python or similar coding languages.
A decade of production software, with Python in the shipped agentic work — is Python + FastAPI microservices end to end — alongside TypeScript/Node and Go as daily drivers. Comfortably 'Python or similar', and a player-coach who reviews production code across all three.
Experience taking AI solutions from conception to launch and architecting AI systems on cloud platforms (e.g., Google Cloud Platform (GCP)).
Conception → launch is the whole job he's done: designed, built and shipped end to end on Gemini 2.5 Flash Lite as the Google × Kaggle capstone; a Claude-powered natural-language CMS taken into a client's production stack; production deploys running on Google Cloud Run; and this site's chat agent on Google's on-device Gemma 4 E2B.
Experience building pipelines for structured and unstructured data using both vector databases and retrieval augmented generation like architectures to power enterprise AI solutions.
RAG and retrieval are in his working stack — embeddings and vector search (pgvector, MeiliSearch), retrieval-augmented pipelines, Google Search grounding at evaluation time in — on top of the structured-and-unstructured data plumbing done the hard way: a HIPAA-grade patient-data pipeline at (2,212 new patients, 9,650 schedules in 2026) and event-driven media-processing queues.
Vector DBs and RAG are real in his builds; the specific enterprise managed services (e.g., Vertex AI Vector Search) are a fast ramp on the craft he already has.
Experience leading technical discovery sessions with customers.
A founder-CIO's daily job for a decade: discovery across five verticals, partnering directly with client CTOs (SkyTracks, Canada) and executives. Turned a 26-year-old export house's institutional knowledge into a specification-first platform by interviewing the principals — then handed them a plain-English AI interface to run it.
Experience architecting AI systems on cloud platforms (e.g., GCP).
Cloud-agnostic architect who deploys, not just diagrams: AI systems on Google Cloud Run, Google ADK and Gemini in shipped builds, plus AWS (ECS, Lambda, DynamoDB, SQS/SNS) and Cloudflare Workers. Architecture as running code, with CI/CD and observability he builds himself.
Preferred qualifications
Master's degree or PhD in AI, Computer Science, or a related technical field.
No graduate degree — saying so plainly rather than dressing it up.
The applied equivalent is public and testable: the Google × Kaggle 5-Day AI Agents Intensive capstone (), Google ADK and MCP in real builds, evals pipelines, and a decade of production systems. The preferred qualification proxies for depth in agentic design — he can demonstrate ReAct loops, hierarchical delegation, state management and tool-calling in running, open-source code, and is happy to be tested on it.
Experience implementing multi-agent systems using frameworks (e.g., LangGraph, CrewAI, Agent Development Kit (ADK)) and complex patterns (e.g., ReAct, self-reflection, hierarchical delegation).
Ships these exact patterns: is a hub-and-spoke A2A mesh — a sequential accuracy-critical parser → six parallel country agents → a synthesis advisor, i.e. hierarchical delegation — with ReAct / plan-execute loops, MCP tool-calling and n8n orchestration with explicit state management. He works in Google ADK; LangGraph/CrewAI specifically aren't his daily tools, but the patterns underneath them are.
Experience in a post-sales or technical consulting delivery function.
A decade of exactly this — customer-facing technical delivery at for clients across Dhaka, Zurich, Oslo, Sandnes, London, Dubai, California, New York and Quebec, owning scoping, delivery and the in-house handover. Post-sales consulting delivery under a founder's title.
Knowledge of large language model native metrics (e.g., tokens/sec, cost-per-request) and techniques for optimizing state management and granular tracing.
Strong on the adjacent half: explicit evals loops, observability he builds himself, n8n state management, and cost-conscious infra by instinct (scale-to-zero Cloud Run at , cost-optimized media queues at ). The honest ramp is the LLM-native specifics — tokens/sec, cost-per-request budgeting and granular tracing as a sustained discipline.
He already optimizes cost and reliability at the infra level and runs evals; the LLM-native metrics and granular tracing layer is a quick, visible ramp on top of habits he already has.
Strengths & honest gaps
An innovator-builder, embedded in customer environments
The role's whole premise — code, debug and jointly ship bespoke agentic solutions inside the customer's stack — is how he already works: a Claude-powered natural-language CMS live in a client's production stack, an embedded engineering pod taken 0→production for Norway's Frontgo, and ' six-agent Gemini mesh. Production-grade reality, not proofs-of-concept, with a founder's high-agency mindset.
Carries frontier AI into legacy reality — modernization that ships
Deliberately comfortable in both worlds: shipping multi-agent meshes and on-device Gemma one day, and pulling a 26-year-old export house onto a modern stack the next — , modernized with a Claude-powered CMS where plain English becomes model-authored PRs and automated edge deploys. He clears the integration complexity, data-readiness and state-management blockers that stop AI reaching enterprise-grade maturity.
Fluent in the exact agentic stack the role names
Multi-agent patterns (ReAct, plan-execute, hierarchical delegation via ' A2A hub-and-spoke), MCP and tool-calling, RAG and vector retrieval (embeddings, pgvector, Google Search grounding), evals and observability, Google ADK — his working vocabulary, in shipped code, not slideware. Google × Kaggle 5-Day AI Agents Intensive capstone done.
A builder with business acumen — startups to enterprise
Accountable for revenue and what ships for a decade as a founder-CIO: scoping, qualifying and pricing engagements, tying field insight to roadmap, and justifying builds by ROI ('s unit economics, 4x faster time-to-market (75% reduction in cycle time) via CI/CD). Has served the full range — founder-led startups through established enterprises — across health-tech, fintech, hospitality, music-tech and export.
A decade on Google's stack — early and ongoing
Not a recent pivot: bet on Flutter early (, a 10,000-user app, 2019), leaned on Firebase in production (notifications, test distribution, a Firestore migration with A/B-tested onboarding for a Dubai client), and today runs many of his latest initiatives on Cloud Run — alongside Gemini, Google ADK, Google Antigravity as his daily dev harness, and on-device Gemma 4 E2B powering this site's chat.
LLM-native performance metrics (tokens/sec, cost-per-request) and granular tracing as a sustained discipline.
Weeks, not quartersHe already runs evals, builds observability, manages agent state in n8n, and optimizes cost at the infra level (scale-to-zero Cloud Run, cost-tuned queues). The LLM-native metrics and granular-tracing layer sits directly on top of those habits — a fast, visible ramp rather than a new muscle.
Google Cloud's post-sales motion and Vertex AI platform specifics at hyperscale.
First quarterThe consulting-delivery and discovery muscle is a decade deep; what's new is Google's scale, GTM context and the Vertex platform surface. He'd close it by learning the existing playbooks before changing anything, and bringing deployment scar tissue rather than a product engineering venture firm's process.
Candid answers
The questions a Google Cloud Consulting hiring manager should ask — answered candidly: why a founder-CIO wants an IC builder seat, how he's both IC and leader, his business acumen and ROI focus, legacy modernization at enterprise scale, Claude versus Gemini/Vertex, the missing graduate degree, and whether Dublin is real.
You co-founded ELO and serve as its CIO — why step into an individual-contributor builder seat?
Because this role removes the one constraint I can't engineer around as a founder. What lights me up is being in the room with a customer — coding, debugging and shipping until AI actually runs in their environment — and this is that, full-time, with frontier Gemini models, Vertex AI and direct access to DeepMind's minds behind it. I've done the management arc and I'm good at it, but I want to be the hands-on innovator-builder embedded in accounts. I'd rather build that from inside Google Cloud than keep consulting from outside it.
Is this an IC role or a leadership role for you — which are you?
Both, and that's the point — I'm a player-coach. The deliverable here is running code, and I'll write it: the Claude CMS and are recent personal builds, not delegated ones. But the JD also asks me to co-build with pre-sales and product teams and instill Google-grade best practices, and that's where the leader half earns its keep — I lift the customer's engineers and the account team without needing an org chart to do it. The leadership multiplies the building; it doesn't replace it.
This role wants measurable ROI, not just clean architecture. Where's your business acumen?
A founder lives or dies by it. For a decade I've scoped, qualified and priced engagements, owned revenue and what ships, and justified every build by the business case — 's unit economics, a 4x faster time-to-market (75% reduction in cycle time), a 26-year-old export house's Canada-first growth. I treat 'production-grade ROI' the way the JD does: the win isn't an elegant diagram, it's AI running in the customer's environment driving a number they care about. I've also run the field-insight-to-roadmap loop my whole career, which is exactly the dual purpose this role describes.
You've worked with smaller clients — can you handle enterprise legacy estates?
I'm deliberately comfortable across the whole range, and legacy modernization is a specialty, not a stretch. I've modernized a 26-year-old export house with a Claude-powered CMS, architected a HIPAA-compliant pipeline and EHR integration at , integrated Vipps payments and financing systems for a Norwegian fintech, and shipped multi-tenant OIDC/RBAC identity — across founder-led startups and established enterprises in five verticals. Carrying a frontier model into a crusty legacy estate, through APIs, data silos and security perimeters, is exactly the bridge this role is about.
Your production AI is Claude-led. This role lives on Gemini and Vertex AI.
I'll be honest rather than retrofit: the AI I've shipped into client production has been Claude-led. But Gemini is genuinely second in my stack, not a gesture — runs six parallel agents on Gemini 2.5 Flash Lite with Google Search grounding and was my Google × Kaggle capstone; this site's chat runs Google's on-device Gemma 4 E2B; I work in Google ADK; my deploys already run on Cloud Run; and I develop daily in Google Antigravity. The craft — multi-agent decomposition, RAG, evals, integration, governance — is model-agnostic, and I'd ramp the Vertex AI specifics visibly.
The posting prefers a Master's or PhD — you have neither.
Correct, and I won't pretend otherwise. My field of study was Computer Science & Engineering at North South University, and I took the shipped-artifact route: the Google × Kaggle Agents Intensive capstone, agentic systems in production, evals pipelines, and nine-plus years of real systems. The preferred qualification proxies for depth in agentic design — ReAct, hierarchical delegation, state management, tool-calling — which I can demonstrate in running, open-source code rather than a thesis. Happy to be tested on it at any depth.
How deep is your RAG and vector-database experience, really?
Real and shipped, and I'll be precise about the edges. I work with embeddings and vector search (pgvector, MeiliSearch), retrieval-augmented patterns, and grounding — grounds against live Google Search at evaluation time — on top of the structured-and-unstructured data plumbing that's the unglamorous half: HIPAA-grade pipelines at , event-driven media queues. What's a fast ramp rather than deep tenure is the specific enterprise managed services like Vertex AI Vector Search — but that's platform surface on top of craft I already have.
The role is in Dublin. Are you genuinely set up to relocate and work in Ireland?
Yes, and I'll keep it honest and low-key. On a personal note, Dublin would bring me a little closer to my sister in the UK, just a short hop away. I'm genuinely intentional about the move — English is my full working language, so there's no ramp on communication, and the Critical Skills Employment Permit is a route I've looked into. I'd be glad to start on Irish hours from day one; the full relocation plan is linked on this page if you want the specifics.
The stack & the builds
The stack behind the verdicts — agentic patterns in shipped code, the data and RAG plumbing, the Google-stack ties old and new, and the builds that prove it.
Editorial export site for Global Jute Trading Ltd — a Dhaka-based sourcing & exporting house with 26 years as a company and 35 years of jute trading under its principals. Built to support their push into deeper international markets, with a particular focus on Canada.
Submitted as the capstone for the Google × Kaggle 5-Day AI Agents Intensive (2026). VisaPros is an agentic visa eligibility advisor utilizing a Google ADK agent-to-agent (A2A) hub-and-spoke architecture — parsing documents once and running six parallel country agents as FastAPI microservices to evaluate fit across destinations.
Strategic health-tech partnership and equity investment in Bangladesh's most advanced clinical practice for wound care, diabetes, and nephrology. Spearheaded high-utility, HIPAA-compliant EHR clinical dashboards and scaled digital patient advocacy.
Coordinated ELO's dedicated engineering pod for Frontgo (Frontpayment's engineering arm) in Norway. The project itself was owned and driven by the client's dedicated project manager; Fauzul's role was to keep the three-engineer pod aligned — setting engineering practices, advising on the Vipps and financing integrations, unblocking communication and requirement gaps, leading hiring/HR, and running the weekly and monthly syncs — through to a clean handover to the in-house team.
Open-source Windows GUI + CLI that links Windows Host directories straight to the WSL 2 Linux filesystem via native directory symlinks. Born the week Google's Antigravity IDE 2.0 broke the WSL Remote workflow — built to restore zero-latency, native-Linux-speed cross-platform development on a Windows desktop.
Frontier AI, shipped inside the customer's real world.
On a personal note, Dublin would bring me a little closer to my sister in the UK, just across the water — and it's where I'd love to embed with customers and build next. I'm genuinely intentional about the move and would be glad to start on Irish hours from day one.
See my Dublin relocation plan →