Research Engineer - Pre-training
Prime Intellect23 days ago
Base Salary
$150k - $350k/yr
Responsibilities
- Build and extend pre-training code in prime-rl, progressing from small experiments to full training runs.
- Evaluate model architecture changes, including MoE and attention designs, based on model quality and training cost.
- Profile training jobs and improve end-to-end throughput through hardware-aware execution.
- Improve training-run resilience through reliable checkpointing and recovery.
- Investigate training dynamics and numerical stability while validating convergence-preserving optimizations.
- Design controlled, reproducible ablations to guide larger experiments.
- Collaborate with data and evaluation researchers on training mixtures, midtraining, and long-context experiments.
Requirements
- Hands-on experience training language models, including modifying training code and analyzing results from actual runs.
- Strong Python skills and experience with PyTorch or a comparable framework.
- Sufficient systems understanding to debug distributed training.
- Practical understanding of model architectures, optimization, and their effects on convergence.
- Ability to design ablations that answer research questions and interpret their limitations.
- Good judgment about independent investigation and collaboration.
Benefits
- Cash compensation ranges from $150,000 to $350,000 and includes equity incentives.
- Flexible work arrangements allow remote work or in-person work at the San Francisco offices.
- Visa sponsorship and relocation assistance are available for international candidates.
- Quarterly team off-sites, hackathons, conferences, and learning opportunities are offered.
Categories
AI ResearchML Engineering
About Prime Intellect
Prime Intellect builds a full-stack platform for training frontier AI models, giving enterprises access to distributed compute and agentic training infrastructure to develop their own systems. The company operates as a platform and open research lab, with customers leveraging its infrastructure rather than building it in-house. Headquartered in San Francisco and privately held, it has raised a Series A round backed by investors including Founders Fund and NVIDIA.