SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 150,000 - 350,000 / annual
Prime Intellect is building an open superintelligence stack—infrastructure that frontier AI labs typically build internally, now available to ambitious AI teams. The company's platform, Lab, unifies compute, environments, evaluations, secure sandboxes, high-performance training, and deployment for post-training at frontier scale, from SFT and RL to tool use, agent workflows, and continuously improving production models. Prime Intellect has raised $150M from Founders Fund, Radical Ventures, NVIDIA, and notable AI/infrastructure operators including Andrej Karpathy, Dwarkesh Patel, and leaders from Ramp, Perplexity, Harvey, Cognition, OpenAI, Databricks, and others.
As a Research Engineer - Pretraining, you will take Prime Intellect's next open models from architecture experiments through full pretraining runs, scaling toward 1–3T parameters. You'll work across the model and training stack, making research decisions and fixing the systems problems that determine whether a run succeeds.
Key responsibilities include:
- Building and extending pretraining code in prime-rl, scaling small experiments into full training runs
- Evaluating model architecture changes (MoE, attention designs) against model quality and training cost
- Profiling training jobs and improving end-to-end throughput in collaboration with systems engineers
- Making long runs resilient to failures through reliable checkpointing and recovery
- Investigating training dynamics and numerical stability to ensure optimizations preserve convergence
- Running controlled ablations to guide larger experiments with reproducible setup and results
- Collaborating with data and evaluation researchers on training mixtures, midtraining, and long-context experiments
The role offers flexible work arrangements with options for remote or in-person work at the San Francisco office, visa sponsorship and relocation assistance, quarterly team off-sites and hackathons, and the opportunity to work on foundational AI infrastructure alongside a mission-driven team.
REQUIREMENTS:
- Hands-on experience training language models, including modifying training code and following effects through actual runs
- Strong Python skills and experience with PyTorch or comparable framework, with systems understanding to debug distributed training
- Practical understanding of model architectures and optimization, including their effect on convergence
- Ability to design ablations that answer research questions and explain what results do and don't show
- Good judgment about what to investigate independently versus when to collaborate with colleagues
