Anthropic
TPU Performance Engineer
$280K–$850K
San Francisco, CA +6 · Remote · Full-time · Posted May 22, 2025
- Location
- San Francisco, CA
New York City, NY
Seattle, WA
New York, New York, United States
Remote-Friendly US (Travel Required)
San Francisco, California, United States
Seattle, Washington, United States - Experience
- Significant accelerator-based ML optimization experience is requested; years of experience correlate with internal job-level requirements.
- Education
- Bachelor’s degree or an equivalent combination of education, training, and/or experience; relevant field of study demonstrated through coursework, training, or professional experience.
- Visa sponsorship
- Visa sponsorship is available, but not for every role or candidate; the company says it will make every reasonable effort to obtain a visa after an offer.
About the job
Optimize performance across ML research, training, and inference systems, especially by designing and tuning TPU kernels and advising researchers on performance impacts of model changes.
What you'll do
- Identify and resolve performance issues across ML systems; design and optimize TPU kernels.
- Provide researchers with feedback on how model changes affect performance.
- Develop performance-focused systems such as low-latency sampling, low-precision inference, and custom communication algorithms.
What they ask for
- Significant experience optimizing ML systems for TPUs, GPUs, or other accelerators.
- Bachelor’s degree or an equivalent combination of education, training, and/or experience.
- Relevant field of study, demonstrated through coursework, training, or professional experience.
Nice to have
- Experience with high-performance, large-scale ML systems, accelerator kernels, or computer architecture.
- Experience with ML framework internals or language modeling with transformers.
Pay
$280,000 — $850,000 USD (annual base salary)
Tools
TPU, GPUs, ML accelerators, transformers
AI focus
Performance engineering for machine-learning research, training, and inference systems, including large language model sampling and inference.
Model-written summary · the original posting is authoritative