Research Scientist Intern, PyTorch On-Device (PhD)

Description

PyTorch is Meta’s deep learning framework for fast, flexible AI/ML experimentation used across industry and backing all of Meta’s ML workloads. More details at https://pytorch.org/

Team scope

• Develop ExecuTorch as the PyTorch solution for on-device AI
• Partner with Reality Labs to run AI on AR/VR hardware
• Partner with Meta family of apps for running AI on iOS and Android
• Partner with hardware vendors on high-performance AI kernels
• Provide on-device AI inference, feature stores, benchmarking, and model delivery

Internship duration

Our internships are twelve (12) to twenty-four (24) weeks long and we have various start dates throughout the year.

Responsibilities

• Develop new or apply existing performance techniques to on-device AI.
• Explore quantization, sparsity, and model/software co-design as solutions.
• Apply knowledge and research to advance the state-of-the-art in on-device machine learning frameworks.
• Collaborate with users and developers of PyTorch and ExecuTorch to enable new use cases inside and outside Meta.

Qualifications

• Currently has, or is in the process of obtaining, a PhD degree in Computer Science or a related STEM field
• Experience in ML compilers, sparsity, quantization, kernel development, or similar as applied to on-device and highly-constrained environments
• Must obtain work authorization in the country of employment at the time of hire, and maintain ongoing work authorization during employment
• Experience working on other AI/ML optimized runtime stacks
• Experience with performance optimization of machine learning models for on-device inference
• Intent to return to degree program after the completion of the internship/co-op
• Proven track record of achieving significant results as demonstrated by grants, fellowships, patents, as well as first-authored publications at leading workshops or conferences such as NeurIPS, MLSys, ASPLOS, PLDI, CGO, PACT, ICML, or similar
• Experience working and communicating cross-functionally in a team environment