Research Scientist / Engineer – Reinforcement Learning Infrastructure

About this role

Luma is seeking a Research Scientist/Engineer to build and scale reinforcement learning infrastructure, enabling their models to become truly useful. This role focuses on designing and implementing distributed systems that handle training, rollout generation, environment execution, and reward computation at a massive scale, utilizing thousands of GPUs. The ideal candidate will have extensive experience post-training LLMs with RL and debugging asynchronous rollout pipelines.

What you'll do

• Design, build, and scale distributed RL post-training systems.
• Build high-throughput rollout generation pipelines.
• Design and build scalable RL environments for agentic tasks.
• Develop reward infrastructure including verifiable rewards and LLM-as-judge pipelines.
• Develop evaluation, monitoring, and debugging tooling.
• Collaborate with researchers to advance training efficiency and stability.

What they're looking for

• Reinforcement Learning (RL)
• Distributed PyTorch
• LLMs
• PPO/GRPO
• RLHF
• vLLM
• SGLang
• NCCL