PhD student position (f/m/d) (Ref. 26/14)

Doctoral research in the WIAS group Data-Driven Mathematical Modeling, led by Clemens Heitzinger, on reinforcement learning and data-driven modelling, covering theoretical foundations and significant applications. Directions include convergence analysis and new reinforcement-learning methods with rigorous off-policy evaluation; reinforcement learning for systems governed by ordinary or partial differential equations, possibly under uncertainty; reasoning and large language models for formal reasoning, theorem proving and hypothesis generation; and learning-based control for automation, robotics and physical AI.

The role includes publishing, collaborating with application partners and helping supervise students.

A master's degree meeting TU Berlin's regulations for mathematics PhD students; a strong background in reinforcement learning, PDEs, optimisation, machine learning or numerical analysis; programming in Julia or a comparable language and in Lean 4 (with Mathlib) and/or Rocq; English communication skills. The part-time appointment (29.25 hours per week on the TVöD Bund scale) is limited to three years, with mobile working possible. Applications go through the WIAS applicant portal, and the advert stays open until the position is filled.