TPU Kernel Engineer

Anthropic; York and North Yorkshire; On-site; GBP 110,000 - 150,000; Full time

Benefits offered by this job

Health insurance

Dental & vision

Parental leave 22 weeks

Flexible PTO

Mental health support

Equity package

Relocation assistance

Home office stipend

Meals in office

Education stipend

Job summary

Anthropic in the United Kingdom seeks a TPU Kernel Engineer to identify and address performance issues across ML systems, including research, training and inference. You will design and optimize kernels for the TPU and provide feedback to researchers on how model changes impact performance.

You will implement low-latency, high-throughput sampling for large language models, adapt models for low-precision inference, and build quantitative models of system performance.

Qualifications

• Bachelor’s degree or equivalent experience required.
• Strong track record solving large-scale systems optimization.
• Experience optimizing ML systems on TPUs, GPUs or accelerators.
• Enjoy pair programming and collaboration.
• Understand accelerators and computer architecture.
• Knowledge of ML framework internals.
• Interest in ML research and societal impact.

Responsibilities

• Identify and address performance issues across ML systems (research, training, inference).
• Design and optimize kernels for the TPU.
• Provide feedback to researchers on model changes and performance impact.
• Implement low-latency, high-throughput sampling for large language models.
• Adapt existing models for low-precision inference.
• Build quantitative models of system performance.
• Design and implement custom collective communication algorithms.
• Debug kernel performance at the assembly level.

Skills

Large-scale systems

Low-level optimization

TPU optimization

GPUs & accelerators

Pair programming

Computer architecture

ML framework internals

Transformers / LMs

Societal impact awareness

Education

Bachelor's degree in related field

Job description

• As a TPU Kernel Engineer, you’ll be responsible for identifying and addressing performance issues across many different ML systems, including research, training, and inference
• A significant portion of this work will involve designing and optimizing kernels for the TPU
• You will also provide feedback to researchers about how model changes impact performance
• Implement low-latency, high-throughput sampling for large language models
• Adapt existing models for low-precision inference
• Build quantitative models of system performance
• Design and implement custom collective communication algorithms
• Debug kernel performance at the assembly level

Benefits

• Comprehensive health, dental, and vision insurance for you and your dependents
• Inclusive fertility benefits via Carrot Fertility
• 22 weeks of paid parental leave
• Flexible paid time off and absence policies
• Mental health support for you and your dependents
• Competitive salary and equity packages
• Optional equity donation matching at a 1:1 ratio, up to 25% of your equity grant
• Retirement plans with competitive matching
• Life and income protection plans
• $500/month flexible wellness and time saver stipend
• Commuter benefits
• Annual education stipend
• Home office stipends
• Relocation support for those moving for Anthropic
• Daily meals and snacks in the office

• Strong candidates will have a track record of solving large-scale systems problems and low-level optimization
• Want to learn more about machine learning research
• Have significant experience optimizing ML systems for TPUs, GPUs, or other accelerators
• Enjoy pair programming (we love to pair!)
• Understanding accelerators at a deep level, e.g. a background in computer architecture
• Are results-oriented, with a bias towards flexibility and impact
• ML framework internals
• Care about the societal impacts of your work
• Strong candidates may also have experience with:
  • Language modeling with transformers
  • Pick up slack, even if it goes outside your job description
  • High performance, large-scale ML systems
  • Designing and implementing kernels for TPUs or other ML accelerators

Education requirements: We require at least a Bachelor’s degree in a related field or equivalent experience