Member of Technical Staff, Kernels
$275k - $550k • San Francisco • FullTime
Posted 2y ago
Job Location
San Francisco
Tech Stack
Remote Work Policy
On-site
Employment Type
FullTime
Categories
Machine Learning Engineer
About the job
Magic is seeking a Kernel Engineer to design, implement, and maintain high-performance kernels that optimize throughput and latency for AI model training and inference. This role is crucial for supporting Magic's long-context windows, which present unique challenges in memory utilization, data movement, and sustained throughput. You will be responsible for the entire lifecycle of kernel development, from design and implementation to deployment and ensuring production reliability.
Responsibilities
- Design and implement kernels for high-performance long-context AI models.
- Own kernel design, implementation, deployment, and production reliability.
- Focus on robustness, extensive testing, and functional correctness while optimizing performance.
- Evaluate porting compute kernels to alternative hardware options.
- Co-design kernels in collaboration with training, inference, and RL teams.
Requirements
- Low-level programming experience targeting AI accelerators (e.g., NVIDIA Blackwell, Google TPUs).
- Experience developing and optimizing GPU kernels in frameworks like NCCL, MSCCLPP, CUTLASS, CuTeDSL, Triton, Quack, Flash-Attention.
- Experience with kernel authoring frameworks such as Pallas/Mosaic (GPU or TPU) or Mojo.
- Deep expertise in computer architecture, low-level machine optimizations, and code generation.
- Breadth of knowledge across Machine Learning.
- Agility, ownership mindset, and grit.
Benefits
- Annual salary range of $275K - $550K.
- Equity as a significant part of total compensation.
- 401(k) plan with 6% salary matching.
- Generous health, dental, and vision insurance for employees and dependents.
- Unlimited paid time off.
- Visa sponsorship and relocation stipend.
- Opportunity to work in a small, fast-paced, highly focused team.