Research Engineer, RL Environments & Synthetic Data (all genders)
Core
Build synthetic environments, task generators, and evaluation harnesses to train and evaluate LLMs on complex, multi-step work involving tools, documents, and code.
Role type
Senior Research Engineer (RL Environments & Synthetic Data)
Builds
Synthetic task environments, data generators, evaluation benchmarks, and QA systems for LLM training.
Domain
AI Research / Large Language Models / Synthetic Data Generation
Deliverable
production ML models
Required skills
Python, containerization (Docker, Kubernetes), async programming, LLM training/evaluation, task design, data analysis, bug debugging, technical writing
Preferred skills
RL training (GRPO/PPO), domain expertise (law, finance, public admin), sandboxing infrastructure, open-source contributions, German language
Technologies
Python, Docker, Firecracker, Kubernetes, EuroHPC, commercial compute
Responsibilities
Design and build environments for long, multi-step work; create generators for thousands of varied tasks; implement tests and rubric-based judges to verify task completion; detect and patch model cheating strategies; measure task set impact on model performance; contribute to open-source releases of environments and benchmarks.
Seniority
Senior, hands-on IC
