Research Engineer Inference
Prime IntellectVisit Prime Intellect website
AI infrastructure company providing an integrated stack for training, evaluating, deploying, and continuously improving agentic models.
Prime Intellect on X (Twitter)Prime Intellect on DiscordPrime Intellect on GitHubPrime Intellect on Documentation
San Francisco, United States
Funding history
Projects
About Prime Intellect
Prime Intellect, Inc. operates AI infrastructure spanning RL environments, hosted training and evaluations, inference, secure sandboxes, and globally sourced GPU compute.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will extend inference-engine integrations, optimize rollout throughput and multi-node MoE inference, reduce model weight-update overhead, evaluate low-precision inference, and partner with researchers on architectures that balance model quality with execution cost.
Requirements
- Experience building or optimizing GPU inference systems
- Understanding of inference-engine internals
- Python skills
- Experience with PyTorch or a comparable ML framework
- Ability to profile end-to-end workloads and verify performance improvements
- Experience debugging numerical or integration issues in ML systems
Responsibilities
- Extend vLLM integration for new model architectures in RL training
- Improve rollout throughput for long-horizon agents
- Optimize multi-node MoE inference
- Reduce trainer-to-inference weight-update overhead
- Build on FP8 and evaluate lower-precision inference
- Use mismatch checks to identify numerical regressions
- Work with researchers on inference-efficient architecture choices
Benefits
- Equity incentives
- Flexible work arrangements
- Remote-work option
- Visa sponsorship
- Relocation assistance
- Quarterly team off-sites
- Hackathons
- Conference attendance