At AMD, the business problem is to drive the performance of post-training workloads on AMD Instinct GPUs, delivering fast, stable, and reproducible training pipelines on ROCm, to accelerate next-generation computing experiences.
Requirements
- Proven GPU performance engineering for deep learning (ROCm/HIP, Triton, or similar).
- Hands-on with SFT. LoRA and RL-based training at scale.
- Strong PyTorch experience (torch.distributed, FSDP/ZeRO or equivalent).
- Proficient in Python and C++; comfortable reading/writing kernels when needed.
- Experience with distributed systems and collective communication libraries.
- Track record of turning profiles into fixes, upstreaming changes, and documenting results.
Responsibilities
- Lead performance for finetuning and RL training solutions on AMD GPUs.
- Improve throughput, memory efficiency, and stability across data, model, and optimizer steps.
- Optimize multi-GPU/multi-node training and communication patterns.
- Contribute efficient kernels/ops and targeted graph-level optimizations.
- Profile, diagnose, and resolve bottlenecks using standard tooling; prevent regressions in CI.
- Ship reproducible pipelines and documentation adopted by internal teams and external developers.
- Collaborate with framework, compiler, and model teams to land durable improvements.
Other
- B.S./M.S./Ph.D. in Computer Science, Computer Engineering, Electrical Engineering, or equivalent
- Location: San Jose, CA preferred. Other US based locations may be considered.
- AMD benefits at a glance.
- AMD and its subsidiaries are equal opportunity, inclusive employers and will consider all applicants without regard to age, ancestry, color, marital status, medical condition, mental or physical disability, national origin, race, religion, political and/or third-party affiliation, sex, pregnancy, sexual orientation, gender identity, military or veteran status, or any other characteristic protected by law.
