Senior RL Infrastructure Engineer - Scaled GPU Systems

Luma; Seattle (WA); On-site; USD 195,000 - 395,000; Full time

Job summary

Luma in Seattle seeks a senior RL systems engineer to design and scale post‑training RL systems, orchestrating trainer, rollout, environment, and reward workloads across thousands of GPUs. You will optimize throughput, stability, and correctness while pushing the frontier of scalable AI.

Responsibilities span building high‑throughput rollout engines (vLLM, SGLang), coordinating weight updates, sandboxed environments, and reward‑model serving; you’ll also develop tooling for evaluation,

Qualifications

• Post-training RL (PPO/GRPO)
• Distributed PyTorch
• RL environments & verifiers
• RL post-training frameworks
• GPU clusters & NCCL/MPI

Responsibilities

• Design, build, and scale distributed RL post-training systems, orchestrating trainer, rollout, environment, and reward workloads across thousands of GPUs.
• Build high-throughput rollout generation, integrating inference engines (vLLM, SGLang), weight synchronization, and asynchronous/off-policy schemes.
• Design RL environments for agentic, multi-step tasks — sandboxed code execution, tool use, computer use, multimodal interaction — scalable to millions of episodes.
• Build reward infrastructure: verifiable/programmatic rewards, reward-model serving, LLM-as-judge pipelines, and defenses against reward hacking.
• Develop the evaluation, monitoring, and debugging tooling that keeps large RL runs stable.
• Advance training efficiency and stability, and turn new post-training ideas into production runs with researchers.

Job description

Luma in Seattle seeks a senior RL systems engineer to design and scale post‑training RL systems, orchestrating trainer, rollout, environment, and reward workloads across thousands of GPUs. You will optimize throughput, stability, and correctness while pushing the frontier of scalable AI.

Responsibilities span building high‑throughput rollout engines (vLLM, SGLang), coordinating weight updates, sandboxed environments, and reward‑model serving; you’ll also develop tooling for evaluation,