Senior Applied Research Engineer

What You’ll Do

The Open Pipe team is building tools to help agents learn from experience, enabling them to perform long tasks autonomously. The work focuses on solving the major bottlenecks that currently prevent agents from self‑improving. Recent releases include ART, the easiest library for RL; RULER, a general‑purpose reward function; and Serverless RL, an API that gives researchers full control over their data, environment, and reward function while outsourcing GPU management.

These initiatives represent systematic progress toward training reliable, self‑improving agents and the role will continue to tackle remaining technical challenges in this space.

This applied research position requires generating and investigating new research ideas to address the remaining obstacles to continuous learning in production. You will collaborate closely with the Open Pipe team to validate theory against real customer tasks and will have access to extensive GPU resources for experimentation. Success in this role depends on fast learning, the ability to ship, and exposure to all layers of our stack, from CUDA kernels to high‑performance LLM tracing dashboards.

• Bachelor’s, Master’s, or PhD in Computer Science, Machine Learning, Robotics, or a related field
• 4+ years of experience in machine learning, or a PhD with 2+ years of relevant industry experience, focusing on model training
• Strong programming skills in Python and hands‑on experience with PyTorch or JAX
• Deep understanding of LLM post‑training techniques, including supervised fine‑tuning, reinforcement learning, and on‑policy distillation
• Experience developing, evaluating, and deploying machine learning models in production environments
• Strong research and problem‑solving skills, with the ability to thrive in ambiguous, fast‑moving settings

• Publications, open‑source contributions, or demonstrable research impact in LLM post‑training or agent learning
• Experience with distributed training, GPU acceleration, and large‑scale model training systems
• Experience leading technically complex projects or mentoring other engineers

We use a mix of custom code and best‑in‑class libraries for building and deploying production services. This includes high‑performance ML frameworks, GPU‑accelerated training pipelines, and cohesive monitoring dashboards to support continuous learning research.