Cognition
Research Engineer, ML Infrastructure
San Francisco · On-site · Full-time · Posted Aug 6, 2026
- Education
- A PhD is one signal among many; demonstrated capability is valued more than credentials.
About the job
Build and operate scalable ML infrastructure that enables researchers to train, evaluate, and improve AI agents efficiently.
What you'll do
- Own distributed training systems, including job launch, checkpointing, fault tolerance, monitoring, and performance optimization.
- Build infrastructure for large-scale coding-agent rollouts, experiment orchestration, and high-throughput data pipelines.
- Debug training failures and implement parallelism strategies to improve reliability and research throughput.
What they ask for
- Deep experience building and operating distributed training systems for large models.
- Strong systems engineering fundamentals in distributed systems, networking, and storage.
- Proficiency in Python and C++, plus systems-level experience with PyTorch or an equivalent deep learning framework.
- Hands-on GPU profiling and memory optimization experience, and experience implementing or optimizing parallelism strategies.
Nice to have
- Track record of building tooling that accelerates research workflows.
- ML knowledge sufficient to engage substantively with researchers.
Tools
Python, C++, PyTorch
AI focus
Build infrastructure for training and evaluating AI agents, including large-scale model training and reinforcement-learning rollouts.
Model-written summary · the original posting is authoritative