Contract & freelance
I ship AI features, agents & evaluation for teams.
For teams with a product in market — I extend it, ship the AI features and agents that keep slipping down your roadmap, and add the evaluation it’s missing. Solo, end-to-end, in weeks. Three of my own products are live and monetized (~$6.5K MRR); that’s the proof, not a deck.
What I build
Four things, done well.
Everything below I’ve shipped in production on my own products first.
AI agents & automation
Multi-source ingest, LLM relevance scoring, and confidence-routed publishing — auto-act, human-review, or reject. The pattern behind Traffic Driver, a 12-platform GTM engine.
Python · FastAPI · GLMLLM evaluation & guardrails
Versioned rubrics with line-level annotations, LLM-as-judge scoring, and Phoenix tracing on every call. Your AI ships measured, not on vibes — the discipline behind Proven Duty.
Phoenix (Arize) · GeminiFull-stack AI engineering
Auth, billing, multi-tenancy, and the parts that slow a team down — shipped onto your existing product. The same work behind three of my own, live and paying today.
Next.js · Supabase · StripeRAG & deep-research pipelines
Multi-step, citation-grounded research that turns raw data into intelligence reports. The engine inside PIE.
Perplexity · SupabaseRates & engagement
Clear rates, no procurement theatre.
Most teams engage me on a fractional retainer — ongoing AI engineering on their product. Day-rate and fixed-scope available too. Outside IR35, remote UK/Europe.
- Day rate£500/day£450–550 · £550 for agent-architecture & eval work
- Week£2,250–2,5005 days · 4+ week booking ~10% commit discount
- Fractional retainer£4,000–8,000/mo2–4 days/week ongoing
- Fixed scopefrom £2,500Outcome-guaranteed · see packages below
Fixed-scope packages
Outcome-guaranteed — you buy a result, not my time.
Discovery + POC agent
1–2 weeksOne use-case, one integration, a working proof of concept.
£2.5–5KAI agent MVP
3–4 weeksSingle use-case shipped: ingest → LLM → routing → review UI.
£8–12KProduction agent system
6–8 weeksMulti-integration, evaluation, guardrails, and observability.
£15–24KLLM evaluation system
2–3 weeksVersioned rubrics, LLM-as-judge, and a dashboard.
£6–9KHow it works
From first message to shipped system.
Scoping call
15–30 minutes. You bring the problem; I bring the questions. Free.
Proposal
A written, fixed-scope proposal within 48 hours — price, timeline, deliverables.
Build
Shipped solo, end-to-end. LLM tracing and evaluation from day one — you see progress, not a black box.
Handover
Code, tests, docs, and a walkthrough. You own it. I’m an email away.
Three products, shipped and monetized — 10,000+ users, ~$6.5K MRR, 194+ tests on one. The repo is the proof.
See the workTerms
The boring (important) bits.
Work with me
Want to build something that ships?
If you've got a product that needs building end-to-end — auth, payments, AI, deployment — let's talk.