Supported ~300 engineers directly as Product lead for PyTorch and AI Platform Pillar Lead (PyTorch, Training, RecSys, Data/TorchArrow). Major projects:- Recommender Systems (https://github.com/pytorch/torchrec): Led the rearchitecture and open-sourcing of Meta's next-gen RecSys stack which allowed us to scale our largest models by ~10x. Adopted immediately by Twitter, Tencent, and many others. Obviously a complex ordeal to get this approved.- Led the rearchitecture of Meta's training stack to modularize and unify infrastructure across monolithic frameworks (ClassyVision, PyText, MMF, Detectron2, internal frameworks, and more).- Led a new AI data loading and processing framework for Meta - a PyTorch-like API on top of flexible underlying compute (CPU, GPU, massive compiled data DAGs). Managed a large cross functional team across AI and data orgs, delivering double-digit % compute savings in our largest AI datacenters.Other areas I supported directly:- PyTorch Core, Distributed, Quantization, Compilers- PyTorch Computer Vision (https://github.com/pytorch/vision)- PyTorch Natural Language Processing (https://github.com/pytorch/text)- PyTorch Speech & Audio Processing (https://github.com/pytorch/audio)- Data Loading (https://github.com/pytorch/data)- Data Processing (https://github.com/facebookresearch/torcharrow)- Composable Functional Transforms (https://github.com/facebookresearch/functorch)

- Tech Lead for Qiskit Aqua, IBM's open-source Quantum Algorithms & Applications Python library- Research in near-term Quantum Algorithms for Quantum Chemistry, Machine Learning, Combinatorial Optimization (https://scholar.google.com/citations?user=WCu0rwwAAAAJ&hl=en)- Technical Lead for U.S. Department of Energy Partnership (Oak Ridge National Lab IBM Q Hub)- Co-organizer, "Workshop on Software for Quantum Applications, Algorithms, and Workflows," with Travis Scholten at IEEE Quantum Week 2020

- Manage new features and own product completeness for Doubleclick Ad Exchange (AdX) buyside, a real-time display advertising exchange auctioning >1M ads per second to >300 external algorithmic bidders- Drive user research and rollout planning across Real-Time-Bidding, AdX Buyer UI, Buyer REST API, Video, Mobile, and Creatives/Policy launches

- Define product requirements and strategic vision for a generalized life sciences data and code platform in collaboration with partners at the Broad Institute of Harvard and MIT- Collected requirements, scoped, and designed the platform’s core data service, built to index and digest life sciences data of arbitrary size and format (e.g. omop, vcf, dicom, mzml), and allow researchers to rapidly understand (<30 seconds) and begin using (<5 minutes) large public datasets

Consumer/Retail Group - Coverage and M&A

Regulatory Calculators Software Engineering Team

Equities Software Engineering Team