Principal Software Engineer – PyTorch Training Frameworks - CA, San Jose, United States
Originally posted 30 January 2026 by the employer.
AMD Instinct accelerators
About the role
AMD is looking for a Principal-level PyTorch training framework expert to help drive performance, scalability, and correctness of large-scale AI training on AMD Instinct accelerators. You will work at the intersection of PyTorch internals, distributed training, and hardware-aware optimization.
What you'll do
- Act as a technical authority for PyTorch training at AMD, setting direction for performance, scalability, and reliability. - Drive optimization of key PyTorch training workloads, including LLMs and foundation models, across single-node and multi-node systems. - Improve and debug training performance in areas such as DDP/FSDP, gradient checkpointing, mixed precision, memory planning, and communication/computation overlap. - Partner with ROCm compiler/runtime, kernel, and driver teams to resolve performance bottlenecks and correctness issues across the full stack. - Contribute to and influence upstream PyTorch design discussions, code contributions, performance fixes, and CI/debug. - Develop and maintain representative training benchmarks, profiling workflows, and performance regression detection for key models.
What you'll need
- Bachelor’s or Master's degree in Computer Science, Computer Engineering, Electrical Engineering, or equivalent. - Deep hands-on experience with PyTorch training, solving systems problems across performance, scaling, memory efficiency, and distributed communication. - Demonstrated technical leadership and cross-team architectural influence driving complex systems problems to execution.
Nice to have
- Deep experience with PyTorch internals and training systems, including Autograd, optimizers, dataloading, compilation paths, and runtime behavior. - Strong distributed training expertise, including DDP, FSDP, tensor/pipeline parallel concepts, collectives (NCCL/RCCL), and multi-node debugging. - Proven track record in performance engineering, including profiling, tracing, kernel/runtime analysis, memory optimization, and scaling studies. - Strong programming skills in Python and C/C++ with the ability to land clean, maintainable changes in large codebases. - Familiarity with PyTorch ecosystem components such as TorchInductor / torch.compile, Triton, CUDA/HIP-style programming models, and performance tooling. - Experience working across OS/hardware boundaries in Linux-based environments, including containers, CI, drivers, and runtimes.
Skills: PyTorch, AI training, AMD Instinct accelerators, distributed training, hardware-aware optimization, LLMs
This role has been open 217 days — well above the 62-day median for Software Engineering Semiconductor roles.
Software Engineering Semiconductor · AI ML Hardware
|
Open roles in category
157
|
Median days open
62 d
|
Median salary
$191k
|
See the full market breakdown ▾Category comparison, skills in demand, and who else is hiring
| Metric | AMD | All employers we track in this specialty (157 roles · 43 employers) |
|---|---|---|
| Open roles in this specialty | 8 | 157 |
| Open roles in the wider Software, Firmware & Systems family | 391 | 6302 · 146 employers |
| Median days open | 62 d | 62 d (same as this employer) |
| Median salary (USD postings) | — | $191k |
Skills observed across this category: PyTorch, AI training, AMD Instinct accelerators, distributed training, hardware-aware optimization, LLMs
Who's hiring in this category
- Micron Technology · 25 open roles · median 40 d
- NVIDIA · 25 open roles · median 224 d
- Qualcomm · 21 open roles · median 200 d
- Infineon Technologies · 12 open roles · median 45 d
- AMD (this employer) · 8 open roles · median 62 d
- Applied Materials · 4 open roles · median 96 d
How we counted: 157 open Software Engineering Semiconductor (AI ML Hardware) roles from 43 employers tracked in the SemiconductorJobs index, counted 26 Oct 2026. Specialty figures count only roles carrying this exact specialty label, so an employer's related work in neighbouring specialties is not included there — it is counted in the wider Software, Firmware & Systems family row. Figures refresh nightly.