Thomas
Rolf
Mannhart

I grew up in Zürich and have been tinkering with computers for as long as I can remember. What started as building mods for games I loved turned into studying Informatics at the University of Zürich, where I spent seven years — first a Bachelor's in Software Systems, then a Master's specializing in Artificial Intelligence. For my bachelor thesis I built a general-purpose range join algorithm for PostgreSQL, which won the UZH Semester Award. For my master's I dove into time series databases and the Kronecker decomposition — an unusual intersection of linear algebra and data compression that I found genuinely fascinating.

Throughout my studies I was always working alongside. My first real software job was at swissbiomechanics, an ETH spin-off, where I single-handedly built a Java application for tracking biomedical analyses and generating clinical reports. That threw me into the deep end of stakeholder communication and independent project ownership at twenty-two. From there I moved to PolygonSoftware, a startup founded by fellow UZH graduates, where I spent three years leading full-stack development and diving into computer vision and machine learning. I designed architectures, supervised dev teams, and talked directly with clients — the kind of small-company work where you end up doing a bit of everything and learning fast.

After finishing my Master's, I joined Ergon Informatik, one of Zürich's most respected software houses, and worked on a time-tracking and workforce planning system for the retail sector. There I sharpened my craft in enterprise Java and Kotlin, handled the full delivery cycle from requirements to support, and found that I really enjoyed mentoring — whether that was onboarding new team members or organizing IT workshops for local students.

These days I'm a Professional AI Engineer at bbv Software Services in Zürich, where I work on the bbv AI Hub — a Swiss-made, model-agnostic enterprise AI platform that's going open source. My day-to-day is building RAG pipelines, agentic workflows, and LLM integrations across enterprise systems. It sits at the exact intersection of my interests: real engineering problems, cutting-edge AI, and the challenge of making complex technology usable and trustworthy for organisations that can't afford to get it wrong.

I also speak and write about AI-augmented software engineering — how AI is changing the way we build software, not just the software itself. I've given talks at the FHNW Alumni Event and bbv webinars on practical methods for integrating AI into the development lifecycle, from requirements analysis through to testing.

My technical toolkit centres on Python, TypeScript, and Java/Kotlin, with deep experience in agent orchestration, RAG, and the Model Context Protocol. I'm fluent in German, English, and French. And when I'm not coding — hot tea, cold beer, good food, thick books, old music, and long boardgame nights.