Summary
Thomas B. Brunner is an AI research engineer based in Germany with nine years of experience bridging reinforcement learning, deep learning, and robotics into production-grade systems. He develops scalable RL solutions for planning and control, currently applying these methods to defense planning problems at Helsing and previously reimplementing AlphaStar-style architectures for efficient experience collection. His master’s work combined Bayesian deep learning and RL to make legged-robot controllers (ANYmal) more uncertainty-aware and safe, including a custom PyTorch package and privileged pre-training pipelines. Earlier roles at Magazino gave him robust production software and embedded systems experience, improving firmware and ROS stacks for deployed robots. He blends academic rigor (MSc with distinction, thesis at ETH Zürich) with hands-on systems engineering, favoring scalable pipelines over toy proofs. Colleagues would note his knack for turning research prototypes into reliable, large-scale RL workflows.
9 years of coding experience
3 years of employment as a software developer
Exchange Program, Exchange Program at Sorbonne Université
B.Sc. Mechanical Engineering, B.Sc. Mechanical Engineering at Technical University Munich
Master's Thesis, Deep Learning, 1.0, Master's Thesis, Deep Learning, 1.0 at ETH Zürich
English, German, Portuguese, French