Research Scientist, AI & Systems Co-design (PhD)

Our teams' mission is to explore, develop and help productionize high performance software & hardware technologies for AI at datacenter scale. We achieve this via concurrent design and optimization of many aspects of the system from models and runtime all the way to the AI hardware, optimizing across compute, network and storage. The team invests significantly into model optimization on existing accelerator systems and guiding the future of models and AI HW at Meta. This drives improved performance, new model architectures and reduces cost of ownership for all key AI services at FB: Recommendations and Generative AI.This is an exciting space that spans exploration and productionization, coupled with close collaborations with industry, academia, Meta's Infrastructure and Product groups. Collaborating closely with product teams, the team's mode of operation is going from ideation and rapid prototyping, all the way to assisting productization of high leverage ideas, working with many partner teams to bring learnings from prototype into production. In addition to the real-world impact on billions of users of the Meta products, our team members have won Best Paper Awards at prestigious conferences such as ISCA, ASPLOS, SOSP, and OSDI, with multiple papers selected for IEEE Micro Top Picks. We regularly publish in ICML, NeurIPS, SC, HPCA, NSDI, VLDB, MLSys, and more. Overall, our work largely corresponds to the research communities of systems in general and especially systems for ML (MLSys, SOSP, OSDI, SIGCOMM, NSDI), hardware architecture (ISCA, ASPLOS), ML (NeurIPS, ICML, ICLR) and supercomputing (SC, ICS).

• Explore, co-design and optimize parallelisms, compute efficiency, distributed training/inference paradigms and algorithms to improve the scalability, efficiency and reliability of inference and large-scale training systems.
• Innovate and co-design novel model architectures for sustained scaling and hardware efficiency during training and inference.
• Benchmark, analyze, model and project the performance of AI workloads against a wide range of what-if scenarios and provide early input to the design of future hardware, models and runtime, giving crucial feedback to the architecture, compiler, kernel, modeling and runtime teams.
• Explore, co-design and productionize model compression techniques such as Quantization, Pruning, Distillation and Sparsity to improve training and inference efficiency.
• Explore, prototype and productionize highly optimized ML kernels to unlock full potential of current and future accelerators for Meta's AI workloads. Open source SOTA implementations as applicable.
• Optimize inference and training communications performance at scale and investigate improvements to algorithms, tooling, and interfaces, working across multiple accelerator types and HPC collective communication libraries such as NCCL, RCCL, UCC and MPI.
• Guide Meta's AI HW requirements and design focusing on performance at System and Silicon levels. Co-design and optimize our AI HW and related software stack for Meta's future workloads, with technology pathfinding and evaluation of cutting-edge, including off-market hardware systems, spanning multi-vendor/generation GPUs and ASICs, including Meta's in-house MTIA.

• Currently has, or is in the process of obtaining a Bachelor's degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience. Degree must be completed prior to joining Meta.
• Currently has, or is in the process of obtaining, a PhD degree in Computer Science, Computer Vision, Generative AI, NLP, relevant technical field, or equivalent practical experience. Degree must be completed prior to joining Meta.
• Specialized experience in one or more of the following areas: Accelerators/GPU architectures, High Performance Computing (HPC), Machine Learning Compilers, Training/Inference ML Systems, Model Compression, Communication Collectives, ML Kernels/Operator optimizations, Machine learning frameworks (e.g. PyTorch) and SW/HW co-design.
• Experience developing AI-System infrastructure or AI algorithms in C/C++ or Python.
• Must obtain work authorization in the country of employment at the time of hire, and maintain ongoing work authorization during employment.

• Experience or knowledge of training/inference of large scale deep learning models.
• Experience or knowledge of either Generative AI models such as LLMs/LDMs or Ranking & Recommendation models such as DLRM or equivalent.
• Experience or knowledge of distributed ML systems and algorithm development.
• Experience or knowledge of at least one of the responsibilities listed in this job posting.