Experience

Where it's worked

• Analyzing company disclosures for regulatory compliance

  Swipe to see the full flow.

  My previous company's main product ran on an engine I built myself, which reviews company disclosures against regulatory requirements to an auditor-ready standard. It extracts content from files accurately, assesses each requirement with guardrails and metrics that confirm quality, traces every finding to its source page with quotes, and pulls out quantitative metrics where they're relevant. It ran over a million times in 2026 and was the basis for all of the company's revenue over the past five years.

• Sustainability data on a global retailer's suppliers

  Swipe to see the full flow.

  A global retailer needed sustainability data on more than 1,500 of its suppliers. I built a system that collected each supplier's documents from the web, pulled out the data, and cleaned and formatted it using more than 50 rules specific to the client. It produced fewer errors than human reviewers and runs in hours, where the same work took a team of four nine months in 2025.

• Tracking regulatory change across dozens of countries

  Swipe to see the full flow.

  A Big 4 consulting firm wanted a daily feed of new and changed regulations from government and regulator websites. I worked out the requirements with the client, then built a system that reads hundreds of websites every day, spots what has changed and pulls out the details. The client got the results as a daily feed.

What I bring

• Technical depth

  I write the code myself, from the document pipelines to the evals that show whether they're working properly. Systems I've built have run in production for paying clients for years.

• Problem solving

  I start by finding the problem most worth solving, then work out what's possible. I won't recommend AI where a simpler solution will do.

• Leadership

  I've led a product organization of up to 18 people and an engineering team of 7, and hired 9 data scientists in one year. As CTO, I took two products from first idea to launch.

• Focus on impact

  I measure the things your team cares about, like accuracy, cost and hours saved, and I test against your own experts' past work so you see the numbers before anything goes live.

• Working with clients

  I've delivered to Big 4 advisory firms, global investors, and a national financial regulator, and I'm used to working between engineers, clients and sales teams.

How I work

• You work with me directly. When a project needs more hands, I bring in engineers I trust and stay responsible for the work.
• Your documents stay under agreed terms. Before anything moves, we agree in writing where documents are stored, which AI providers can see them, and how long anything is kept. It can run on your own infrastructure.
• You own what's built for you. The custom code, prompts, and evals for your project are yours.
• The right model for the job. I pick the AI model that fits the task and the budget, and I test new ones before switching when something better comes out.

Talk it through