onsite
Principal Software Engineer, Performance Tooling - Microsoft Corporation
Software Engineer
Lead the design and implementation of high‑performance tooling for AI frameworks, optimizing execution across diverse hardware from supercomputers to edge devices, while collaborating with hardware and software teams to push the limits of AI performance.
About the role
Key Responsibilities
- Architect and develop performance‑critical components for AI frameworks, ensuring efficient execution on CPUs, GPUs, and specialized AI accelerators.
- Collaborate with hardware teams to integrate low‑level optimizations, including CUDA kernels and memory‑bandwidth tuning.
- Design and maintain profiling, debugging, and benchmarking tools that provide actionable insights to developers and researchers.
- Lead code reviews, mentor junior engineers, and drive best practices in performance engineering and software quality.
- Work cross‑functionally with AI researchers, product managers, and platform teams to translate algorithmic advances into production‑ready solutions.
Requirements
- 10+ years of software engineering experience with a strong focus on performance and systems programming.
- Expertise in C++ and Python, with deep knowledge of CUDA and GPU programming models.
- Proven track record of optimizing large‑scale AI workloads and delivering measurable speedups.
- Excellent problem‑solving skills, ability to work in a fast‑paced, collaborative environment.
- Strong communication skills and experience mentoring engineering teams.