Summary
Jett Janiak is a technology specialist focused on AI safety at the European Commission with nine years of hands-on research and engineering experience spanning mechanistic interpretability, LLM evaluation, and real-world reinforcement learning. He holds a mathematics Master’s with distinction from the University of Warsaw and has authored and co-authored multiple arXiv and conference works on LLM residual streams, chain-of-thought faithfulness, and small-transformer circuits. Jett’s background bridges rigorous research (ML Alignment & Theory Scholars, LASR Labs, AI Safety Camp) with production-facing systems engineering—building DRL pipelines and Rust simulators for city traffic and fixing critical bugs in a 50 PB Ceph cluster. He combines deep theoretical insight with practical delivery, from designing sim2real DRL extensions to shipping training libraries for small language models. Based in Brussels, he brings a rare mix of formal math training and proven impact across both academic and operational AI safety domains.
10 years of coding experience
3 years of employment as a software developer
Master's degree, Mathematics, 5.0 with distinction, Master's degree, Mathematics, 5.0 with distinction at University of Warsaw