Software Engineering Manager, Collective Communications

In this role, you will be a member of the Host Side Collective Communication Libraries (HCCL) software team and part of the Host Side Networking (HSN) organization within the DC Networking and Network Infra organizations. The team develops and owns the software stack around collective communication libraries for the internally developed Meta AI accelerator, Meta Training and Inference Accelerator (MTIA). At the high level, the team's charter is to enable and performance optimize HCCL, our internally developed communication library for the MTIA. Currently, one of the team’s focus areas is standing up our first communications stack targeting various AI training workloads. The team is building the communications software stack including hardware customized features, software benchmarks and performance tooling. Specifically we are looking for people to lead designing, developing and operating some of the largest AI infrastructure in the world. This is a rare opportunity to work with the leading AI experts in the industry and build cutting edge AI communications infrastructure.


Help define technical roadmap for the team, drive execution of associated tasks and support the team in resolving dependencies Guide and mentor team members to develop appropriate skillsets to grow in their careers, and where necessary address under performance Collaborate effectively with other groups across the wider organisation such as Co-Design, Software, Hardware, Infrastructure and Operations teams Communicate cross-functionally and drive engineering efforts Interact with external partners as needed in resolving dependencies associated with objectives


8+ years, or PhD + 4 years, of software engineering work experience, including hands on technical management 2+ years of experience managing a networking related Software Engineering Team BS or MS in Computer Science or related technical discipline or equivalent experience Working knowledge of Collective Communications Libraries such as NVIDIA Collective Communications Library (NCCL) and Message Passing Interface (MPI) Experience with software development for Distributed and Embedded systems Demonstrated experience recruiting, building, structuring, and leading technical organizations, including performance management Experience supporting, coaching, mentoring, and developing software engineers Experience with distributed GPU reliability/performance improvement on RoCE/Infiniband Knowledge of network transport stack such as RoCE (RDMA) Experience working with Deep Learning frameworks like PyTorch, Caffe2 or TensorFlow Knowledge of Machine Learning, Deep Learning and Large Language Models