Member of Technical Staff, Performance Engineering
Physical Superintelligence · Boston
Frontier labremote
At a glance
Highlights
- Own a workload's time end to end to make one cluster produce more physics.
- Build AI systems to discover and commercialize transformative physics breakthroughs.
- Measure real cluster jobs to guide queue policy and capacity decisions.
What they offer
- Competitive salary
- Benefits
- Meaningful early-stage equity
Ideal candidate
Whole-stack performance owner who lives in measurements and traces, fixes slow systems in their own repos, and cares about turning cluster time into real physics discovery.
Requirements
- 3+ years whole-stack performance work with before-and-after traces spanning multiple layers
- 3+ years systems programming reading, modifying and verifying C++, CUDA or Triton code
- Scale expertise in one area: multi-node AI training, inference serving at scale, or HPC simulation at scale
- Works from measurements and profilers, reasoning about memory hierarchy, occupancy and roofline
- remote policy
- Hybrid, 3 days per week in person in Cambridge, MA
- visa sponsorship
- No visa sponsorship; must be authorized to work in the US
Lab profile
About Physical Superintelligence
Building AI systems to discover and commercialize transformative physics breakthroughs at scale.
What they build
- AI systems for physics discovery
- Data center performance optimization
Known for
- AI agents that learn physics
- Verifiable physics research campaigns
- AI/HPC infrastructure at scale
- Thermal, fluid and power systems modeling
Posted · Last seen