Member of Technical Staff - Pre-Training Evals Infra (Inference)
Reflection AI · Global Office
Frontier labonsite
At a glance
Highlights
- Own inference performance and correctness for pre-training evaluations.
- Turn one-off tuning and debugging into reliable, scalable systems.
- Improve autoscaling and fleet efficiency for evaluation workloads.
What they offer
- Top-tier compensation with salary and equity
- Stock options for all contributors
- Comprehensive medical, dental, vision and life insurance plus wellness allowance
- Lunch and dinner provided daily in office
- 22 weeks paid parental leave
- Unlimited PTO in the U.S. and 30 days in the U.K.
- Regular off-sites, happy hours and team celebrations
Ideal candidate
A rigorous, ownership-driven systems engineer who loves deep performance debugging in distributed systems and partnering with researchers to turn one-off tuning into reliable infrastructure for open models.
Requirements
- Track record of improving performance or reliability in complex distributed systems and inference infrastructure
- Understand CPU execution, GPU architecture, memory hierarchy and interconnects and how they shape performance
- Understand inference including prefill and decode, batching, attention, KV-cache management and distributed execution
- Approach optimization through careful measurement, hypothesis, experiments and bottleneck analysis
- Bring rigor to correctness, distinguishing implementation errors, numerical differences and experimental noise
- Take ownership of ambiguous problems and work effectively with researchers and engineers
- visa sponsorship
- Visa sponsorship available
Lab profile
About Reflection AI
Research lab making intelligence open and accessible for everyone to use, customize, and build on.
What they build
- Open foundation models
- Aligned general agents
- Enterprise AI deployments
Known for
- Frontier model pre-training
- Post-training and alignment
- Distributed training infrastructure
- Secure cloud platform foundations
Posted · Last seen