Research Engineer, RL Frontiers

What you'll do

• Own the post-training stack end to end and train models using proprietary simulators and environments.
• Build RL environments with engineering teams, using simulation, formal tools, tests, and hardware measurements for rewards.
• Create custom data mixes and benchmarks that improve model performance and avoid reward hacking.

What they ask for

• Hands-on post-training experience, including SFT and RL (e.g., PPO, GRPO), and experience shipping models people used.
• Experience improving a model on a task it was not built for; ideally with scarce data.
• Strong PyTorch and distributed-training engineering skills.
• Understanding of RL systems and how to make the training loop faster and cheaper.

Nice to have

• Experience with RL or post-training frameworks (e.g., verl, OpenRLHF, TRL, prime-rl) or inference engines (vLLM, SGLang).
• Experience in agentic, multi-turn RL with verifiable rewards, distillation, specialized-model training, ASIC design or verification, accelerator kernels, or large-scale compute infrastructure.

Pay

$400K+ USD base salary

Tools

SFT, PPO, GRPO, PyTorch, verl, OpenRLHF, TRL, prime-rl

AI focus

Post-train frontier models with supervised fine-tuning and reinforcement learning to automate kernel development, ASIC engineering, production pipelines, and 3-D hardware operating systems.