Frontier RL Inference Engineer (Remote)

Prime Intellect, Inc.; San Francisco (CA); Remote; USD 150,000 - 350,000; Full time

Benefits offered by this job

Visa sponsorship

Relocation assistance

Flexible remote work

Quarterly team off-sites

Job summary

Prime Intellect, Inc. is building open frontier AI infrastructure. You will own the inference stack for RL research, extending vLLM integrations and optimizing multi-node MoE inference on NVL72 clusters to enable large-scale, efficient RL training.

Join a mission-driven lab in San Francisco offering visa sponsorship, relocation assistance, and flexible remote options. You will collaborate with researchers to balance model quality against execution cost while advancing frontier AI.

Qualifications

• Experience building or optimizing GPU inference systems.
• Strong Python skills and PyTorch experience.
• Ability to profile end-to-end workloads and translate measurements into improvements.
• Experience debugging numerical or integration issues in ML systems.

Responsibilities

• Extend prime-rl’s vLLM integration to bring new model architectures into RL training, with small-scale validation before larger runs.
• Improve rollout throughput for long-horizon agents, tuning concurrency and KV-cache reuse as context lengths grow.
• Optimize multi-node MoE inference, including expert parallelism and the tradeoffs of separating prefill from decode.
• Reduce trainer-to-inference weight-update overhead through sharding-aware transfer and communication overlap, while keeping rollout workers on the intended policy version.
• Build on FP8 and evaluate lower precision. Use trainer/sampler mismatch checks to catch numerical regressions and measure the effect on RL training.
• Work with model researchers on architecture choices that improve inference efficiency, balancing model quality against execution cost on the target hardware.

Skills

Python

PyTorch

Profiling

ML systems

vLLM

Tools

vLLM

Job description

Prime Intellect, Inc. is building open frontier AI infrastructure. You will own the inference stack for RL research, extending vLLM integrations and optimizing multi-node MoE inference on NVL72 clusters to enable large-scale, efficient RL training.

Join a mission-driven lab in San Francisco offering visa sponsorship, relocation assistance, and flexible remote options. You will collaborate with researchers to balance model quality against execution cost while advancing frontier AI.