Eleven years of engineering, three of them spent making AI behave.

I’m Jaunius Pinelis, a software engineer from Lithuania. I studied Computer Science at the University of Edinburgh and started my career in Cambridge, maintaining systems that had been alive far longer than my patience for them — which turned out to be the best possible training for enterprise work.

Since then I’ve built regtech products, government forest-management systems, banking microservices, and — for the last three years — generative AI platforms. At Danske Bank’s GenAI tribe I designed the assistant platform and LLM proxy that internal developers used to reach Azure OpenAI, vector search and speech-to-text, and built incident-classification pipelines on SageMaker. At Flo Health I shipped agents on Google ADK, stood up LLM-judge evaluation in Databricks, and advised on AI procurement.

Alongside that I’ve taught continuously since 2020 — .NET, JavaScript, Python, generative AI, agentic coding — and now lead AI engineering content and cohorts at Turing College. Teaching keeps me honest: you can’t hand-wave an architecture to thirty engineers who will implement it on Monday.

The consultancy is deliberately small. You get me, not a pyramid.