Research Engineer – Inference

Prime Intellect

. Build inference systems powering Prime Intellect’s RL research, scaling toward 1–3T parameter models on NVL72 clusters .

Posted 10/16/2026full-timeSan Francisco • California • United StatesMid-LevelSenior💰 $150,000 - $350,000 per yearWebsite

Core Competencies

Role fit

Core Competencies

Use this summary to align your resume positioning with the role.

Demonstrates expertise in building and optimizing GPU inference systems, with strong proficiency in Python and experience with PyTorch or comparable ML frameworks. Capable of profiling workloads and debugging numerical issues to enhance performance in reinforcement learning environments.

Highest-signal resume keywords

GPU Inference Systems OptimizationPython ProgrammingPyTorch Framework ExperienceEnd-to-End Workload ProfilingNumerical Debugging in ML Systems

ATS Keywords

Tailor your resume

Applicant Tracking System Keywords

Tip: use these terms in your resume and cover letter to boost ATS matches.

Hard Skills

Inference System DevelopmentPerformance Improvement TechniquesConcurrency TuningKV-Cache Reuse OptimizationMulti-Node MoE InferenceFP8 EvaluationTrainer/Sampler Mismatch ChecksSharding-Aware TransferCUDA Kernel DevelopmentTriton Kernel Work

Tools & Technologies

VLLM IntegrationML FrameworksInference Engines

Industry Keywords

Reinforcement LearningModel ArchitectureInference EfficiencyHardware Execution Cost

Tech Stack

Tools & technologies

Node.jsPythonPyTorch

About the role

Key responsibilities & impact

Build inference systems powering Prime Intellect’s RL research, scaling toward 1–3T parameter models on NVL72 clusters
Extend prime-rl’s vLLM integration to bring new model architectures into RL training
Validate integrations at small scale before larger runs
Improve rollout throughput for long-horizon agents by tuning concurrency and KV-cache reuse
Optimize multi-node MoE inference, including expert parallelism and separating prefill from decode
Reduce trainer-to-inference weight-update overhead through sharding-aware transfer and communication overlap
Keep rollout workers on the intended policy version
Build on FP8 and evaluate lower precision
Use trainer/sampler mismatch checks to identify numerical regressions and measure effects on RL training
Collaborate with model researchers on architecture choices that improve inference efficiency while balancing model quality and hardware execution cost

Requirements

What you’ll need

Experience building or optimizing GPU inference systems, with an understanding of inference-engine internals
Strong Python skills
Experience with PyTorch or a comparable ML framework
Ability to profile end-to-end workloads and turn measurements into verified performance improvements
Experience debugging numerical or integration issues in ML systems
Contributions to vLLM or another inference engine (nice to have)
CUDA or Triton kernel work, particularly for low-precision inference (nice to have)

Benefits

Comp & perks

Equity incentives
Flexible work arrangements
Option to work remotely or in-person at offices in San Francisco
Visa sponsorship
Relocation assistance for international candidates
Quarterly team off-sites
Hackathons
Conferences and learning opportunities