Leonard T.

I take AI systems from prototype to production : RAG, agents, LLM pipelines and forecasting models that hold up under real traffic and real money. Most teams can build a demo. Shipping one that survives contact with users is a different job, and it's the one I do. WHAT I'M USUALLY BROUGHT IN FOR • GenAI to production : RAG and agent architecture, evaluation harnesses, LLM pipelines, cost and latency work. Turning a prototype into something you can put in front of customers. • Forecasting and time-series : demand, price, load, capacity. Feature and validation design, and putting a model where a decision actually depends on it. • Private / on-prem LLM : self-hosted models and RAG over internal data, when the data can't leave your perimeter. WHY ME RATHER THAN SOMEONE CHEAPER The rigour: a PhD in machine learning applied to European day-ahead power markets (a domain where a bad forecast loses money by lunchtime) and the Frontier Prize for best paper at IDA 2023. The track record: 14 months of applied research on diffusion models for a US client, and three full applications designed, built and launched solo with AI agents as the build team, the third live in a single day (see the portfolio). HOW I WORK Long-term hourly, async-friendly across US and EU time zones. Happy to take a 30-minute call and tell you honestly whether I'm the right person for it.