Software Engineer, Systems ML

Job Title

Software Engineer, Systems ML

Role Summary

Join the MTIA (Meta Training & Inference Accelerator) Software Tooling team to design and build developer tooling for Meta's in-house AI accelerator ASICs. The team provides debugging, profiling, memory analysis, monitoring, and reliability tooling across the MTIA stack.

The role focuses on debugging, sanitizer technology, fault isolation, and profiling for AI workloads running on MTIA hardware, working across compiler, runtime, kernel, and hardware layers to deliver production-quality tooling.

Experience Level

Mid-level. The posting specifies 4+ years of experience in systems software engineering, performance engineering, developer tooling, or closely related fields.

Responsibilities

Design, implement, and maintain tooling that enables debug, profiling, and reliability analysis for accelerator workloads.

• Lead design and development of profiling and debugging tooling: performance profiling, live debugging, and core dump support for accelerator workloads.
• Build kernel-level diagnostics, multi-rank fault isolation, and performance analysis capabilities.
• Own significant components end-to-end: design, implementation, production rollout, and maintenance.
• Collaborate with compiler, runtime, kernel, and hardware teams to add instrumentation, hooks, and sanitizers into the software stack.
• Develop AI-native tooling approaches and automation, including LLM-guided diagnostics where appropriate.
• Work with internal product teams to prioritize tooling investments based on developer needs.
• Document architectural decisions and share debugging methodology and best practices across teams.

Requirements

Key technical must-haves and useful optional skills drawn from the posting.

Must-have:

• 4+ years of systems software, performance engineering, developer tooling, or closely related experience.
• Experience building debugging, sanitizer, profiling, or diagnostic tools for complex software/hardware systems.
• Proficiency in C++ and Python, including low-level systems programming and scripting for tool automation.
• Experience across multiple layers of the stack (compiler, runtime, OS/driver, hardware) and tracing issues across those boundaries.
• Proven track record of leading technical design and delivering tooling or infrastructure to production.

Nice-to-have / Preferred:

• Experience with accelerator ecosystems (GPU/CUDA, TPU, custom ASICs) and related toolchains (cuda-gdb, nsight-compute, nsight-systems).
• Familiarity with ML framework internals (PyTorch execution, torch.compile) and AI compiler stacks (MLIR, LLVM, TVM, Triton).
• Linux kernel and driver-level debugging, sanitizer and dynamic-analysis technology, and runtime shadow-memory techniques.
• Experience with Linux profiling/debugging tools (gdb, perf, eBPF, ftrace), hardware performance counters, ELF/DWARF metadata.
• History of building developer tools adopted by large engineering teams or contributions to relevant open-source projects.

Education Requirements

Minimum: Bachelor's degree in Computer Science, Computer Engineering, or a relevant technical field, or equivalent practical experience. Preferred combinations: Bachelor's + 5+ years, Master's + 3+ years, or PhD + 2+ years in systems software, developer tooling, or accelerator software development. References to formal degrees and equivalence were present in the posting.

About the Company

Company: Meta Platforms

Headquarters: Menlo Park, California, United States

American technology company that develops social networking products (Facebook, Instagram, WhatsApp) and invests in virtual/augmented reality hardware and software through Reality Labs, focusing on connectivity, advertising, and immersive computing experiences.