Job Details

GPU Performance Engineer at Yandex. Focuses on optimizing GPU utilization and performance for key Yandex services. Involves hypothesis formation, optimization, profiling, and tool development.

We manage one of the company's most scarce and expensive resources — Graphics Processing Units (GPUs). Their efficient use is critically important for the operation of Yandex's key services. Our mission is to ensure maximum return and effect from each GPU card. What tasks await you: • Improving GPU Utilization Efficiency Formulate hypotheses and research ways to improve GPU utilization efficiency, participate in the implementation and adoption of the most profitable solutions. Formulate recommendations and best practices for performance improvement. • Optimization and Profiling Identify performance bottlenecks and eliminate them using profilers, optimize memory access, latency, and throughput. • Development of Diagnostic Tools Create and improve tools for quick identification and elimination of infrastructure problems that affect utilization efficiency, stability, and speed of GPU computations. • Research and Implementation of Modern Solutions Study the latest approaches to organizing infrastructure for training and inference, evaluate their effectiveness, and implement them in projects. • Architecture Analysis, Testing, Integration Interact with developers, ML engineers, and system architects. Participate in evaluating hardware solutions and suggest improvements for future GPU generations.

We expect you to: • Know Python and have experience in systems programming • Have worked with the PyTorch framework • Have optimized GPU application performance and improved GPU utilization efficiency • Have worked with GPUs (NVIDIA) and CUDA • Apply parallelization approaches for distributed inference or training It will be a plus if you: • Are proficient in C/C++ or similar low-level languages • Have worked with RL training libraries for LLMs

Yandex is a community. There are sports clubs, a book club, and an esports community here. These are not all the bonuses — the full list is here: https://yandex.ru/jobs/pages/benefits?utm_campaign=ya_nanimaet