Research Scientist / Engineer – Reinforcement Learning Infrastructure

The role

You will build and scale distributed reinforcement learning post-training systems. This involves orchestrating trainer, rollout, environment, and reward workloads across thousands of GPUs.

What you will do

• Design, build, and scale distributed RL post-training systems.
• Build high-throughput rollout generation integrating inference engines.
• Design RL environments for agentic, multi-step tasks.
• Build reward infrastructure including reward-model serving and LLM-as-judge pipelines.
• Develop evaluation, monitoring, and debugging tooling.
• Advance training efficiency and stability with researchers.

Required qualifications

• Hands-on experience post-training LLMs with RL (PPO/GRPO-family, RLHF, RLVR) at meaningful scale.
• Extensive distributed PyTorch training and parallelism (FSDP, Tensor/Pipeline/Expert Parallel) for foundation models.
• Experience building RL environments, reward functions, verifiers, or evaluation harnesses for LLM agents.
• Deep familiarity with RL post-training frameworks (veRL, OpenRLHF, TRL, Ray orchestration) and rollout inference engines (vLLM, SGLang).
• Strong understanding of GPU clusters, networking, and communication libraries (NCCL, MPI) under mixed training and inference workloads.

Nice to have

• Running RL training across 100+ GPUs, including asynchronous or disaggregated trainer/rollout architectures.
• Containerization and orchestration (Kubernetes, Ray) for large environment fleets and sandboxed workloads.
• Research contributions in RL for LLMs, or open-source contributions to RL training frameworks.

About the company

Luma's mission is to build unified general intelligence that can generate, understand, and operate in the physical world. They believe multimodality is critical for intelligence.