Join Microsoft's pioneering AI Frameworks team, instrumental in developing cutting-edge AI software that powers the world's most advanced AI models. This role offers a unique opportunity to contribute to the software stacks for next-generation supercomputers and the innovative Maia-300 AI accelerator, working collaboratively with hardware teams and partners. You will engage directly with ML researchers and developers to optimize and scale model training and inference, operating at the nexus of AI innovation, bespoke AI hardware, and sophisticated systems.

The team thrives on a cross-disciplinary, collaborative, and inclusive culture. As a key member, you will be involved in designing system software, developer tools, and enhancing the overall developer experience to facilitate large-scale model training and inferencing on novel AI hardware. An entrepreneurial spirit, proactive initiative, and the ability to drive progress rapidly are essential for this position.

Key responsibilities include defining project requirements, scoping solutions, estimating effort, and managing deliverable schedules. You will apply rigorous engineering principles to architect robust and maintainable systems. Collaboration is paramount, involving close work with hardware designers, system engineers, tool developers, performance analysts, and ML application developers.

Core duties involve hands-on software development using languages such as C/C++, C#, and Python, adapting as project needs dictate. This role demands a proven track record of solving complex technical challenges and a drive to push the boundaries in building comprehensive AI stacks.

We are seeking candidates with a Bachelor of Science or an advanced degree in computer engineering, computer science, or a related field, coupled with over 8 years of extensive software development experience, including leading at least two commercialized features from inception to deployment. A minimum of 5 years of experience with C/C++ and Python, particularly in low-level development, is required.

Familiarity with NVIDIA/AMD development environments, including CUDA and ROCm for Deep Neural Network training and inference, is essential. A strong foundation in software engineering principles and architecture design, along with deep technical understanding, is crucial. Candidates should possess strong intellectual curiosity, a passion for learning new technologies, and experience developing software/firmware and system software for multicore chips.

Excellent cross-team collaboration skills and a genuine desire to work within a team of researchers and developers are highly valued. Strong written and oral communication abilities are also a key requirement for this role. This position is designed for experienced professionals who can take initiative and tackle the most challenging problems in the AI ecosystem.