Applied Research - Sovereign AI
$150k - $300k • Remote • San Francisco • FullTime
Posted 1d ago
About the job
Prime Intellect is developing an open superintelligence stack, providing infrastructure for AI labs. Their platform, Lab, integrates compute, environments, evaluations, secure sandboxes, and high-performance training for post-training at scale, including SFT, RL, tool use, agent workflows, and continuous model improvement. They are building open-source frontier AI models for long-horizon tasks and the platform their research team uses. The role focuses on building post-training and agent systems for national institutions, public-sector organizations, and regulated industries, addressing needs for models and agents that operate within strict boundaries for data, deployment, access, auditability, and human authority, while enabling self-improvement without external model dependency.
Responsibilities
- Build private improvement loops that transform sensitive data into training signals, evaluations, and deployable model updates within data boundaries.
- Train domain-specific agents with permissioned, traceable actions, bounded tools, and evaluated against real procedures.
- Develop high-assurance evaluations for reliability, calibration, robustness, human escalation, and compliance under adversarial conditions.
- Own end-to-end improvement loops from trace selection to deployed updates within operator boundaries.
- Create agents with explicit permissions, evidence trails, bounded tools, and human checkpoints.
- Design evaluations for reliability, calibration, recovery, and compliance with real procedures.
- Conduct experiments on how privacy, access, and infrastructure constraints affect learning, developing transferable methods for high-assurance settings.
Requirements
- Experience in RL/post-training, private ML systems, high-assurance evaluation, agent safety, regulated-domain ML, public-sector technical work, or deployment under strict constraints.
- Hands-on experience building and debugging model-training, post-training, agent, or evaluation loops.
- Strong experimental judgment with meaningful baselines and clear failure attribution.
- Proficiency in Python and ability to build runnable, inspectable, and extensible artifacts.
- Comfort owning ambiguous problems and communicating progress to partners and teammates.
Benefits
- Cash Compensation Range of $150-300k
- Equity incentives
- Flexible Work (San Francisco, New York, London, or remote)
- Visa Sponsorship & relocation support
- Professional Development budget
- Team Off-sites & conference attendance