Deep Learning Compiler Engineer

Overview In this role you will contribute to CUDA Tile, a tile-based programming model in CUDA 13.1, by designing compiler transformations and lowering passes to optimize tile-based kernels across NVIDIA GPUs. You will work with MLIR-based dialects, define public APIs, and drive performance improvements in a fast-paced, product-focused team. This position offers the chance to impact deep learning workloads and the broader GPU software stack. You’ll join a collaborative, innovative environment that values technical excellence and practical solutions. Compensation / Benefitsbase salary range 152,000 USD - 241,500 USDequitycomprehensive benefits packageremote work optionscareer growth opportunities ResponsibilitiesDesign and implement compiler transformations for CUDA TileDevelop MLIR-based dialects and lowering passesOptimize tile-based kernel performance across multiple GPU generationsDefine public APIs and contribute to compiler/optimization techniquesEngage in general software engineering tasks including testing and debugging Key requirementsBachelors, Masters or Ph.D. in Computer Science, Computer Engineering or related field (or equivalent)3+ years in compiler optimization, performance analysis and IR designStrong C/C++ programming, debugging, performance analysis and test designAbility to work independently and lead development effortsStrong interpersonal skills for a dynamic, product-oriented teamstrong collaborationself-motivationeffective communicationMLIRLLVMXLA