01 / 08
GESTURE
GALAXY
A cinematic procedural galaxy that lets players fly through 180,000 stars, enter solar systems and discover planets—with full hand-gesture piloting performed locally through a webcam.
AI Product Engineer
Products × Agents × Systems
London, UK · 2026
I BUILD AI THAT SURVIVES REALITY.
Technical Solutions Manager by role. AI product builder by practice. I ship original products and dependable operational systems—from spatial interfaces to the infrastructure that makes them real.
↓Exploreselected work
01 / SELECTED WORKEight systems · two public · six anonymised.
GESTURE
02 / HOW I BUILDThe stack is standard. The operating discipline isn’t.
THE SHIP
KIT.
ONE PRACTICAL SPINE
I use a deliberately compact stack to move from a proven interface to a secure, integrated production system—without losing control of data, deployment or rollback.
REACT
TypeScript · Vite · Next.js
Product UI and full-stack application delivery
SUPABASE
Postgres · Auth · RLS
Schemas, policies, migrations and access control
N8N
Self-hosted · L2 certified
Official UK Ambassador · production workflows
VERCEL
Preview · release · rollback
Repeatable production delivery and release gating
CLOUDFLARE
DNS · TLS · subdomains
Traffic, certificates and controlled public exposure
THE ENGINEERING SURFACE
MORE THAN
MORE THAN
A WEB STACK.
The same product owner can move through application code, database policy, integration logic, infrastructure and the operational handoff.
React, Next.js, component systems, state and end-to-end application delivery.
RLS policies, migrations, reconciliation, single-source and single-writer controls.
n8n workflows, scheduled syncs, REST integrations and failure-aware pipelines.
VPS administration, regional deployment, Cloudflare DNS and certificate handling.
Data tooling, local processing, MediaPipe, LLM workflows and agentic delivery.
Shopify, Amazon, finance, logistics, analytics and workspace platforms.
THE ACTUAL DIFFERENTIATOR
FRAMEWORKS GET BUILT.
FRAMEWORKS GET BUILT.
DISCIPLINE GETS SHIPPED.
Local interface and mock data first. Prove the product before introducing cloud complexity.
RLS, migrations, one writer per production database and read-only access by default.
One-way where possible, scheduled, monitored and checked for drift.
Merge, deploy and data writes are individually gated—not bundled into one irreversible move.
Pause the old system instead of deleting it; preserve a rollback path under one hour.
Release, observe, resolve and improve from real use. Shipping is the start of the evidence.
PLATFORM FLUENCY
SHOPIFY PLUS · AMAZON · ERP · XERO · HUBSPOT · POLAR · AIRTABLE · GOOGLE WORKSPACE · SLACK
THE DIFFERENCE / 01MORE THAN SOFTWARE DELIVERY
!
WHEN
REALITY
FOUGHT BACK.
The best engineering decisions appeared after contact with real constraints—not inside the initial specification.
- Re-architected deployment for a restricted network
- Kept private files entirely on-device
- Removed premium code physically from the free build
BREAK→UNDERSTAND→REBUILD BETTER
03 / CAPABILITIESDepth across the whole delivery surface.
WHAT
I BRING
TOGETHER.
ONE OWNER / FOUR DISCIPLINES
The value is in the connections: product judgement informing architecture, data informing UX, and automation surviving production.
Product engineering
Discovery, requirements, UX systems, full-stack implementation, authentication, testing, release and iteration.
React · TypeScript · Node · Vite · VitestAI systems
Agentic workflows, knowledge retrieval, evaluation-minded automation, training and human-in-the-loop operating models.
Claude · LLM APIs · RAG · prompt systemsData + integration
Schema design, migrations, API orchestration, ERP and CRM integration, access control and reliable sources of truth.
Postgres · Supabase · RLS · REST · webhooksPlatform + automation
Deployment pipelines, self-hosted infrastructure, background workflows, observability and production incident ownership.
Vercel · n8n · Linux · Cloudflare · Docker04 / AI-NATIVE DELIVERYHUMAN
HUMAN
JUDGEMENT.
MACHINE
LEVERAGE.
PARALLEL BUILD CAPACITY
HUMAN DIRECTION
×
LEVERAGE
LEVERAGE
AI EXPANDS THE POSSIBLE.
ACCOUNTABILITY STAYS HUMAN.
I use agents as a disciplined delivery multiplier: parallel investigation, implementation and verification under one product and technical direction. Priorities, judgement, validation and responsibility remain mine.
Agents accelerate research, build and test loops
Human ownership governs decisions and production
SPECIFYDELEGATEVERIFYSHIP
05 / HOW I WORKFrom ambiguity to dependable operation.
FOUR
MOVES.
Fast does not mean careless. It means reducing uncertainty in the right order.
FIND THE REAL CONSTRAINT
Map the people, process, data and failure modes before choosing the shape of the solution.
FRAMEDESIGN FOR REALITY
Choose the smallest system that can survive its users, integrations, permissions and operating context.
DESIGNSHIP THE WHOLE LOOP
Build, migrate, test and deploy—not just the interface. Keep the release path repeatable.
SHIPSTAY FOR THE TRUTH
Watch real use, resolve incidents, measure adoption and evolve the system from evidence.
OWN06 / WORK INDEXCapability-led. Employer-safe.
THE
RANGE.
No matching work. Try another search.
07 / ABOUTThe profile behind the work.
THE SHORT VERSION
BUILDER.
BUILDER.
OPERATOR.
TRANSLATOR.
“My best work happens where the problem is cross-functional, the system is messy, and somebody needs to own the whole answer.”
I make ambiguity buildable.
I turn conversations, workflows and scattered requirements into a clear product and technical direction.
I build beyond the interface.
Data models, permissions, integrations, migrations, deployment and operational controls are part of the product.
I use leverage deliberately.
AI agents accelerate the work, while evidence, review and human judgement keep it trustworthy.
I stay accountable.
I care whether the system works on Monday morning for the person whose job depends on it.
THE THROUGH-LINE IS SIMPLE.
I TURN COMPLEXITY INTO USEFUL SYSTEMS.
08 / LET’S TALKLondon · open to ambitious AI product roles.
CAPABILITY.
RELIABILITY.
HUMAN JUDGEMENT.