Product Manager - AI/HPC for MTIA chips and systems to accelerate AI at Meta.

Product Lead for Google's core ML software infrastructure (compilers, runtimes). Impacting users and partners across Google (DeepMind, Cloud, Ads, Search, YouTube, etc.) and directly influencing the roadmap for ~100 engineers.- Delivered (0->n) a standardized ML runtime infrastructure for scalable and performant ML execution across HW (TPU, GPU, CPU) and frameworks (JAX, PyTorch, Tensorflow). This is the default stack at Google and many industry partners (open source OpenXLA).- Proposed and initiated (0->1) a cross-organizational program to bridge Google's internal and cloud ML infrastructure, translating to significant cloud revenue and infrastructure efficiency through foundational products, tooling and processes.

Managed Azure's market-leading product portfolio of HPC compute instances to deliver supercomputing HW+SW infrastructure. Drive the product strategy, roadmap and execution to productize complex infrastructure technologies end-to-end; from silicon to global-scale datacenters. Additionally, managed the software platform and ecosystem: services, tools, libraries for virtualization, performance and usability. Hands-on from technology evaluation, system design, go-to-market strategy to customer/partner sales engagements. Incubated strategic initiatives to invest in novel products and services. Present to executives (CXOs, VPs) in EBCs and public forums.Media interview (The Next Platform); invited presentation (Ansys); 1 conference publication (Super Computing).

Delivered platform capabilities for the virtualization, performance and scalability of cloud computing instances. Demonstrated unprecedented scalability of distributed HPC applications on the cloud, 20x over competition, consequently critical to winning multiple deals.Founded roadmap for the HPC & AI software stack on Azure: optimized OS images, container images, extensions, and recipes. Delivered automated provisioning of GPU and InfiniBand (networking) enabled compute with O(10^5) MAUs and >99.9% SLA.

Leader of the Scientific Computing interest group with >400 members world-wide. Created a technical competency assessment and development program to measure and improve mandated KPIs. Forged relationships in academia and industry for research, recruitment and training.Program Lead for a next-gen HPC multi-physics simulation platform. Delivered a multi-year development roadmap to business stakeholders of a $xB product line to strengthen technology offering and increase market share. Implemented the architecture (MPI, OpenMP, C++11/14), mathematical model and delivered PoC which was strategically built for HPC in the cloud."Be Outstanding - Out of the Ordinary Award" for technical expertise impacting business.1 patent memo (ML for scientific computing), 3 conference publications.

Delivered unprecedented 4x performance gains in a legacy multi-physics simulation workflow by leveraging heterogeneous parallel computing. First HPC (MPI, OpenMP, CUDA) deployment on the cloud (Azure, Google) for a petrotechnical application (in 2014) resulting in a scalable, premium product offering to clients with direct annual license revenue of $xM.SME (subject matter expert) advising energy industry clients on optimal and safe productionof assets by leveraging multi-physics modeling and simulation (completion, stimulation).1 patent pending with USPTO.Journal reviewer (Physics of Fluids). Internal publications on HPC for energy industry applications.

US Navy Submarine ­'Crashback': Enabled the first­ large scale simulations (LES, 20M cells) of a submarine propeller with a hull and analyzed the complex flow in a mission-critical condition. Code (MPI, Fortran) is now used at the US Naval Surface Warfare Center to design stealth submarines.Turbulence modeling in Large Eddy Simulation (LES): Developed a constrained optimization-based, hybrid turbulence model (RANS-LES) to enable inexpensive yet statistically resolved flow simulations in complex geometries replacing inaccurate, parameterized models.3 competitive university awards for excellence in academics.4 journal publications, 7 conference publications in prestigious, high impact-factor journals.

­Taught computational structural analysis (FEA) and solid mechanics to senior and junior year undergraduate students.

Performed numerical optimization (Matlab) of energy functionals to determine optimal shapes for internally strained elastic thin films.

Developed novel k-d tree based upwind meshless numerical method (C++) to simulate inviscid flow past multi-element airfoils.