I help startups and scaleups build agentic AI and LLM systems that work in production. I'm often the first senior AI hire a company has, on a flexible basis. I help founders make the right technical calls early, build the core system, and then hand over to an in-house team when the time is right.

Production agentic AI systems

Building a demo with an LLM takes an afternoon. Building an agent you can trust with real users, real data and real consequences is a different job.

I design and build:

• Agentic workflows that use tools, make decisions and carry out multi-step tasks, with guardrails and human oversight where it matters
• RAG systems that answer questions from your own documents, with sources
• Structured extraction that turns unstructured text, documents or speech into validated data
• LLM integrations embedded in your existing product

I take reliability seriously. That means typed, validated outputs, proper evaluation, observability, and clear behaviour when things go wrong. My core stack is Python, FastAPI and PydanticAI across AWS, GCP and Azure.

Deployed where you need it. Most systems run in the cloud. Some can't, or shouldn't. That might be because of sensitive data, regulation, inference costs, latency or working offline. In those cases I deploy on-premise GPU servers or on-device models using open-weight LLMs, speech recognition and embedding models. This gives you full control of your data and costs without giving up capability.

AI strategy and technical due diligence

Early technical decisions are expensive to undo. I help founders work out what's realistic, what to build first, and whether to build, buy or use an existing model. I also carry out technical due diligence for investors assessing AI startups, and help founders prepare for it.

Typical questions I help with:

• What can AI realistically do for our product?
• Should we build our own, buy, or use an existing model?
• Cloud, on-premise or on-device, and what are the trade-offs?
• Do we have the right data, skills and infrastructure?
• What should we build first, how long will it take, and what will it cost?

Team building and mentoring

When you're ready to build an in-house AI team, I can help you hire, assess candidates and mentor your first data scientists and ML engineers. That way the system I've built is in good hands.

Who I work with

Mostly startups and scaleups, with specialist experience in:

• Health tech and life sciences: clinical NLP, medical speech recognition, patient-facing conversational AI and predictive modelling for drug discovery
• Defence tech: AI for constrained, disconnected and high-assurance environments, including local speech and language models for field use
• Other sectors: legal, logistics, retail and e-commerce