I build and ship AI systems inside hedge funds and buy-side firms — agentic workflows, research copilots, and production pipelines that survive contact with a trading floor. 15+ years across Elliott Management, LMR Partners, Credit Suisse, Daiwa Capital Markets.
A focused 4–6 week engagement to design, prototype and ship a production-ready AI system inside your firm. Research copilots, earnings pipelines, reconciliation agents, risk workflows.
Most firms adopt Claude Code or Cursor without a plan — and quietly plateau. I work with engineering teams on the full stack: data boundaries and security posture, skills libraries and repo conventions, and the evaluation habits that turn ad-hoc AI usage into a repeatable team workflow. Delivered as workshops, hands-on pairing, and written playbooks.
Ongoing advisory retainers or one-off expert sessions. I help teams navigate vendor selection, build-vs-buy decisions, and the specific failure modes of LLMs in regulated environments.
A small selection of recent projects. Production systems at previous employers are described at the boundary of what I can share publicly.
Co-founder and CTO of a SaaS club management platform. Designed the full stack end-to-end — auth, membership, bookings, billing — and took it from sketch to paying customers while holding a day job.
A multi-stage agentic pipeline that ingests earnings calls and SEC filings, extracts structured facts, and produces analyst-grade briefs. Built with rigorous evaluation — 75/75 facts verified, zero hallucinations.
Led development of real-time risk and P&L systems at LMR Partners and Elliott Management. VaR engines, OMS integration, Pre/Post-trade processing and Compliance workflows, distributed compute — the unglamorous infrastructure that keeps a multi-billion dollar fund running.
A personal research project exploring agentic patterns for systematic trading. Specialist agents — analysts, fundamental signal detection, portfolio constructors, risk overseers — coordinate through a shared state and message bus. Built to stress-test what production-grade agentic systems look like outside a toy demo: explicit roles, bounded context, deterministic guardrails around non-deterministic components, and evaluation harnesses for each agent in isolation and in concert.
I write about building AI systems for serious production environments — financial services in particular, but the principles travel. Evaluation discipline, architecture patterns, and the gap between LLM demos and systems that actually hold.
I've spent my career at the intersection of engineering and financial markets — building the systems that price, manage risk, and reconcile real capital at serious firms.
My thesis for this decade is simple: AI is about to reshape how the buy-side works, but only where it's built with the same rigour we apply to any other production system. That's the work I want to do, and the bar I hold myself to.
I'm based in London, BEng Computing from Imperial College. Outside work I play squash, and drink more specialty coffee than is probably advisable.
If you're leading an AI initiative at a hedge fund, buy-side firm or applied-AI company — or weighing whether to start one — I'd like to hear about it. The best fit is usually a 30-minute conversation first.