Now

Through June 2026 I led AI extraction for HEOR Systematic Literature Reviews at Bridge Medical Consulting (a Red Nucleus company), running the work as two parallel tracks.

Track 1: ship what works. Laid the engineering foundations under the existing Jupyter-notebook workflow — automating the manual steps, containerizing it, and exposing it through an interactive UI so the whole team could use it, not just one or two specialists. Cut turnaround roughly in half.

Track 2: design what comes next. A structured extraction approach for the ~100 interdependent fields per study: smart ingestion, staged validation, agentic escalation on low-confidence fields, and audit trails as a first-class output.

Both tracks shared one constraint: a domain reviewer needed to look at any extracted value and see, in seconds, where it came from and how confident the system was. That was the bar everything else was measured against.

Open to new AI engineering roles and collaborations — production AI systems that domain experts can audit, not just trust. The fastest way to reach me is email.

Tinkering with a meditation AI assistant, exploring how technology can support mindfulness and personal growth. It’s a different kind of challenge from my day-to-day work, and I find the intersection of AI and wellbeing genuinely interesting.

Two threads at the moment. On the AI side, LLM evaluation and the tooling around it: reading the research and following how practitioners like Hamel Husain and Shreya Shankar think about measuring LLM systems in production. On the engineering side, designing reliable ML and AI systems from the software fundamentals up, and exploring end-to-end development of modern full-stack applications through the University of Helsinki’s Full Stack Open.