Research Scientist
Applied Compute1 year ago
Base Salary
$200k - $335k/yr
Responsibilities
- Design and optimize RL training and inference pipelines across large GPU clusters.
- Build tooling and observability for inspecting, profiling, and debugging training runs.
- Implement reliable and efficient systems accounting for low-precision numerics and distributed-training edge cases.
- Partner with researchers to bring frontier post-training capabilities into production deployments.
Requirements
- Experience programming with and managing training jobs on large-scale GPU systems.
- Curiosity and ability to understand all levels of the training stack.
- Bias toward fast implementation with strong reliability and efficiency.
- Familiarity with open-weights model architecture and inference.
- Background in reinforcement learning or integrating inference with RL training loops.
- Preferred experience with PyTorch, JAX, DeepSpeed, high-performance computing, large-scale clusters, open-source ML infrastructure, or technically creative published and side projects.
Benefits
- U.S. base salary range of $200,000–$335,000 plus equity and comprehensive benefits.
- 100% employer-paid health insurance for the employee and family.
- Unlimited PTO and paid parental leave.
- Daily lunches and dinners.
- Transportation and relocation support.
- 5% 401(k) match.
- Based in San Francisco with work from the Mission office.
Tech Stack
Categories
About Applied Compute
Applied Compute builds enterprise AI systems that turn internal knowledge into custom models and deployable agent workforces. Its platform and forward-deployed teams create evals, train models on proprietary data, and run continually learning agents inside customer environments. The company is privately held and headquartered in San Francisco, serving enterprises that want in-house, domain-specific AI rather than generic foundation models.