Senior GPU Compiler Development Engineer

About the Role

We seek a seasoned compiler engineer to join our GPU Assembler team focused on crafting cutting-edge tools that enable optimal performance for applications in high-performance computing (HPC), deep learning (DL), and autonomous driving sectors. The role centers around developing industry-leading Assembler and Code Analysis Tools to help globally dispersed HPC developers maximize efficiency on current and upcoming NVIDIA GPU architectures. The team delivers essential platform components accelerating general-purpose GPU computation, including the widely used NVDISASM disassembler for CUDA binaries and performance-critical assembler modules.

Key Responsibilities

• Lead the architecture and implementation of NVIDIA's GPU compute Assembler and Disassembler tools.
• Design and develop binary analysis and instrumentation capabilities such as call graph generation, GPU program register usage analysis, and binary patching.
• Collaborate closely with GPU architecture and debugger/profiler teams to gather requirements, provide new features, and enhance product quality.
• Coordinate with teams building complementary components for seamless integration, reliability, and robust code generation.
• Engage with customers and partners to collect feedback, fostering innovation and feature enhancement.

Required Qualifications

• Bachelor’s or Master’s degree in Computer Science, Computer Engineering, or allied disciplines.
• Over 8 years of experience in low-level systems software development, with at least 6 years focused on compilers, assemblers, binary utilities, or tools for binary analysis.
• Strong proficiency in C/C++ programming and solid analytical skills.
• Experience in any core area of compiler development such as feature support, code generation, or compiler infrastructures.
• Good knowledge of assembly language, processor instruction set architectures (GPU ISA knowledge is advantageous), and processor architecture principles.
• Familiarity with object file formats like ELF and debugging formats such as DWARF.

Preferences That Will Make You Stand Out

• Expertise in debugger, profiler, binary tools, or linker internals, with experience in binary analysis or instrumentation tools like BOLT.
• Regular and effective utilization of AI tools (e.g., Cursor, Codex, Claude) to boost productivity.
• Understanding of GPU programming and compute APIs, including CUDA and OpenCL.