May 2022 – Aug 2022 · 3 mos · Menlo Park, California, United States

• I worked as part of PyTorch Perf Infra team who focuses on infrastructure and data-driven optimizations for PyTorch. I worked as core maintainer in PyTorch open-sourced Github repository. My project aimed to support inter-op parallelism in Static Runtime CPU inference Engine. I developed infrastructure for inter-op parallelism for PyTorch models in CPU runtime giving up to 90% improvement in performance of PyTorch models and merged into open-sourced GitHub PyTorch repository. I also designed and developed framework for execution of asynchronous operations on custom executors. This enabled the runtime to delegate the asynchronous execution of sub-graphs on parallel threads improving throughput of data-centers. The project contributions are merged into the Pytorch Github Repository

Mar 2021 – Aug 2021 · 5 mos · Bengaluru, Karnataka, India

• Machine Learning Systems @Samsung.
• Research and Development for the acceleration of deep neural network use-cases on edge devices via distributed heterogeneous computing, performance, and memory optimizations for AI-enabled Samsung camera vision use-cases.

Feb 2019 – Mar 2021 · 2 yrs 1 mo · Bengaluru, Karnataka, India

• Artificial intelligence for vision usecases @samsung
• Responsible for development of system software for machine learning use-cases on Samsung mobile devices.
• Heterogeneous inferencing of deep learning frameworks on mobile devices
• Power efficient and accelerated inference of complex deep learning solutions on mobile devices

Jun 2017 – Feb 2019 · 1 yr 8 mos · Bengaluru, Karnataka, India

• Responsibility includes the Acceleration of deep neural computer vision use-cases for Samsung mobile devices.
• Work lies in the intersection of High-performance computing and AI on different DNN Engines like Caffe, Armnn, Tensorflow, Snapdragon Neural Processing Engine, SNPE in common grounds of heterogeneous computing units like CPU, Graphics Processing Unit, Digital Signal Processors, and Neural Processors.
• Contributions for over 20x performance/power/memory improvement and efficient deployment of over 20 camera USP features. Represented Samsung at organizational level at Device Solutions Research, Samsung Korea for the integration of camera vision use-cases on Neural Processing Unit for S10/Note10 devices.

Jun 2016 – Jul 2016 · 1 mo · bangalore

• Internship in Samsung IOT AI Research and Development
• Real-Time Video Streaming by android devices and Onvif camera to the server performing vision intelligence of object detection and classification.
• Development of an Android application responsible for sending real-time video frames via RTMP protocol to the server. Deployment of Nginx RTMP media server on Amazon AWS for incoming video frames and provide HLS video output after DNN inference.

May 2015 – Aug 2015 · 3 mos · Delhi

• Android application development internship for Lifestyle Application named Stollup.
• Responsibilities: front-end application development, database management, and server interaction, low-level designing for the application.
• The application was successfully deployed on Google Play Store with over 9k+ downloads.
• Play Store Link - https://play.google.com/store/apps/details?id=in.strollup.android