Member of Technical Staff - ML Infra
Trajectory · San Francisco
Infrastructureonsite
At a glance
Highlights
- Own the innermost loop of a continual learning platform for large agentic models.
- Drive 10x monthly gains across scale, throughput, memory efficiency and latency.
- Build self-improving loops where agents propose optimizations and validate gains.
Ideal candidate
A systems-minded engineer with deep ownership in training, inference or kernels who loves squeezing measurable performance from demanding systems and partnering with researchers to turn ideas into reliable gains.
Requirements
- 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 with built systems or measurable optimizations
- Strong Python skills plus specialty languages such as C++, CUDA or Triton
- Understanding of PyTorch, JAX or comparable framework internals with strong profiling and debugging skills
- Ownership and clear communication, working closely with researchers to deliver measurable, correct and reliable gains
Lab profile
About Trajectory
Builds a continual-learning platform helping production models learn from real workflows.
What they build
- Continual-learning platform
- Post-training stack for LLMs and agents
- SDK and API surfaces
Known for
- Continual learning from production telemetry
- Custom LLM and agent training and serving
- Novel continual-learning algorithms
- ML training and inference infrastructure optimization
Posted · Last seen