August 27
🏢 In-office - Vancouver
• The d-Matrix compiler team is looking for exceptional candidates to help develop the compiler backend - specifically the problem of assigning hardware resources in a spatial architecture to execute low level instructions. • The successful candidate will be motivated, capable of solving algorithmic compiler problems and interested in learning intricate details of the underlining hardware and software architectures. • The successful candidate will join a team of experienced compiler developers, which will be guiding the candidate for a quick ramp up in the compiler infrastructure, in order to attack the important problem of mapping low level instructions to hardware resources. • We have opportunities specifically in the following areas: • Model partitioning (pipelined, tensor, model and data parallelism), tiling, resource allocation, memory management, scheduling and optimization (for latency, bandwidth and throughput).
• Bachelor's degree in Computer Science with 7+ Yrs of relevant industry experience • MSCS Preferred with 5+ yrs of relevant industry experience • Ability to deliver production quality code in modern C++. • Experience in modern compiler infrastructures, for example: LLVM, MLIR. • Experience in machine learning frameworks and interfaces, for example: ONNX, TensorFlow and PyTorch. • Experience in production compiler development. • Algorithm design ability, from high level conceptual design to actual implementation. • Experience with relevant Open Source ML projects like Torch-MLIR, ONNX-MLIR, Caffe, TVM. • Passionate about thriving in a fast-paced and dynamic startup culture.
Apply Now