Machine Learning Performance Engineer - CUDA Python
Job title: Machine Learning Performance Engineer - CUDA Python in USA at MatchPoint Solutions
Company: MatchPoint Solutions
Job description: MatchPoint Solutions is a fast-growing, young, energetic global IT-Engineering services company with clients across the US. We provide technology solutions to various clients like Uber, Robinhood, Netflix, Airbnb, Google, Sephora, and more! More recently, we have expanded to working internationally in Canada, China, Ireland, UK, Brazil, and India. Through our culture of innovation, we inspire, build, and deliver business results, from idea to outcome. We keep our clients on the cutting edge of the latest technologies and provide solutions by using industry-specific best practices and expertise.We are excited to be continuously expanding our team. If you are interested in this position, please send over your updated resume. We look forward to hearing from you!Pay $70 to 80/hrMachine Learning Performance Engineer - CUDA PythonDuration: 6-month contract with the likelihood to extendLocation: Remote but candidates must be willing to travel to different customer sites.*Must be willing to travel*Must have strong pre-sales abilities i.e. presentation skills, communication skills, etc.*Must be willing to help train employees and customersYour part here is optimizing the performance of our models both training and inference. We care about efficient large-scale training, low-latency inference in real-time systems, and high-throughput inference in research. Part of this is improving straightforward CUDA, but the interesting part needs a whole-systems approach, including storage systems, networking, and host- and GPU-level considerations. Zooming in, we also want to ensure our platform makes sense even at the lowest level is all that throughput actually goodpu88080/hr 80/hrt? Does loading that vector from the L2 cache really take that long?
- An understanding of modern ML techniques and toolsets
- The experience and systems knowledge required to debug a training run's performance end to end
- Low-level GPU knowledge of PTX, SASS, warps, cooperative groups, Tensor Cores, and the memory hierarchy
- Debugging and optimization experience using tools like CUDA GDB, NSight Systems, NSight Compute
- Library knowledge of Triton, CUTLASS, CUB, Thrust, cuDNN, and cuBLAS
- Intuition about the latency and throughput characteristics of CUDA graph launch, tensor core arithmetic, warp-level synchronization, and asynchronous memory loads
- Background in Infiniband, RoCE, GPUDirect, PXN, rail optimization, and NVLink, and how to use these networking technologies to link up GPU clusters
- An understanding of the collective algorithms supporting distributed GPU training in NCCL or MPI
- An inventive approach and the willingness to ask hard questions about whether we're taking the right approaches and using the right tools