Trajectory
Member of Technical Staff - ML Infra
San Francisco · On-site · Staff · Full-time · Posted Oct 6, 2026
About the job
Develop and optimize Trajectory’s ML training, inference, and GPU-kernel infrastructure, improving performance and enabling continual learning workflows.
What you'll do
- Build and optimize distributed training and RL infrastructure, including scheduling, parallelism, checkpointing, and recovery.
- Optimize production and rollout inference, including batching, scheduling, and KV-cache management.
- Develop benchmarks and observability; optimize GPU kernels and runtimes, and validate performance improvements with researchers.
What they ask for
- Strong fundamentals in distributed systems, networking, storage, and failure recovery, with experience shipping demanding systems.
- Deep specialization in training infrastructure, inference systems, or GPU kernels, demonstrated through systems built or measurable optimizations.
- Strong Python skills and relevant specialty languages such as C++, CUDA, or Triton.
- Understanding of PyTorch, JAX, or comparable framework internals, plus strong profiling and debugging skills.
Nice to have
- Hands-on experience with training and RL stacks such as Miles, SkyRL, Prime Intellect’s verifiers, or comparable systems.
- Experience with vLLM, SGLang, collective communication, or ML compilers.
Tools
Python, C++, CUDA, Triton, PyTorch, JAX, Miles, SkyRL
AI focus
Build continual-learning infrastructure agents that propose optimizations, run experiments, measure gains, and learn from results across training, inference, and kernels.
Model-written summary · the original posting is authoritative