Most teams don't need another deck about AI. They need someone who's wired up LLMs in production and knows where the real failure modes live. That's where I come in.

• LLM application architecture: choosing the right shape for the problem
• Retrieval-augmented generation (RAG) systems that actually retrieve the right thing
• Agentic workflows and tool-use design that stay debuggable as they grow
• Evals, observability and regression testing for non-deterministic systems
• Model selection and cost control: matching capability to budget
• Inference infrastructure, caching strategies and latency tuning

I work hands-on in AI/ML engineering day-to-day at a leading research lab, so the practices I bring aren't second-hand. They're what I use myself.

I'm also the creator of Footstep, a practical demonstration of shipping production AI end-to-end. If you want to see the kind of engineering I bring to client work, Footstep is the proof.

Flexible engagement models depending on what you need:

• Advisory. Architecture reviews, technical due diligence, a second pair of eyes on your AI roadmap.
• Fractional. Embedded with your team part-time to lead AI/ML engineering without the full-time hire.
• Build with your team. Hands-on pairing on the hard parts (evals, RAG quality, agent reliability), levelling up your engineers as we go.