Remote RL Infrastructure Engineer for Frontier Training

OpenKrill; United States; Hybrid; USD 150,000 - 350,000; Full time

Benefits offered by this job

Cash Compensation Range of $150-350k +

Remote or SF office

Visa sponsorship & relocation for intl

Quarterly off-sites & learning

Opportunity to work with a mission-led

Job summary

Prime Intellect is building the open frontier AI stack to accelerate research and production at scale. Our platform Lab unifies compute, environments, evaluations, sandboxes, and deployment for frontier-scale models.

Join a fast-moving team that structures RL and distributed training workloads, optimizes performance end-to-end, and contributes to open-source tooling used across the AI community. Remote-friendly with in-person offices in San Francisco; visa sponsorship is available.

Qualifications

• Experience engineering AI/ML infra for training or inference.
• Proficient in PyTorch and distributed training tools.
• Experience optimizing across kernels, memory, and communication.
• Experience with data/tensor/pipeline parallelism.
• Strong GPU profilers and debugging skills.
• Ability to operate with high ownership in a fast-moving environment.

Responsibilities

• Build and optimize systems infrastructure for large-scale RL and distributed training.
• Improve end-to-end training efficiency across compute, memory, networking, and scheduling.
• Design low-level performance optimizations including kernels and runtimes.
• Work on distributed training across data, tensor, and pipeline parallel workloads.
• Shape architecture of RL training stack, including async rollout and post-training systems.
• Contribute to open-source libraries and internal infra.
• Collaborate with researchers and engineers to translate bottlenecks into improvements.
• Stay at frontier of training/inference systems and hardware-aware optimization.

Skills

Large-scale AI/ML infra

PyTorch expert

Distributed training frameworks

GPU architecture & profiling

Performance optimization

Data, tensor, and pipeline parallelism

Tools

PyTorch

DeepSpeed

Megatron-LM

FSDP

vLLM

Ray

CUDA

Triton

Job description

Prime Intellect is building the open frontier AI stack to accelerate research and production at scale. Our platform Lab unifies compute, environments, evaluations, sandboxes, and deployment for frontier-scale models.

Join a fast-moving team that structures RL and distributed training workloads, optimizes performance end-to-end, and contributes to open-source tooling used across the AI community. Remote-friendly with in-person offices in San Francisco; visa sponsorship is available.