NVIDIAimage

2021-Present

I am currently employed as a Senior Software Engineer at NVIDIA. I work with Maxine and Broadcast products. For the algorithms and features in the products, I work on AI-based video-enhancement features such as webcam-denoising, artifact-reduction and, generative AI-based image animation such as eye-contact and live-portrait models.

I also focus on bringing these models to production using NVIDIA Cloud Functions (NVCF) and NVIDIA Inference Microservices (NIMs). In doing so, I have also built and release AI Foundation models, complete with Model Card++. NVCF endpoints are cloud-based inference endpoints that are supported by Triton Inference Servers.

I also focus on efficient enterprise-level ML platforms within NVIDIA, with a particular focus on MLops for computer vision teams. I worked with multiple-teams and architected cross-cutting infrastructure tools for data governance, management, visualization and preparation. I also built and migrated workloads for training and benchmarking of various computer vision models including architecting systems for logging, visualization, KPI tracking, continual training, hyper-parameter optimization artifact and configuration management etc.

For obvious reasons, some of the work that was performed while being employed by NVIDIA will not be made public here. Others that are already launched or made public are listed below.

Tenure Satistics:

  • Patents filed: 1

  • 2024-Present
    Maxine NVIDIA Inference Microservices (NIMs).
  • 2022-Present
    NVIDIA AI Foundations and NVIDIA Cloud Functions.
    • Live-portrait.
    • Live-Portrait + Voice Font.
  • 2022-Present
    Generative AI Models
    • Eye-contact.
    • Live-portrait.
    • Speech-Live Portrait.
  • 2022-2023
    MLops

    Architecting systems for and migrating computer vision workloads to Maglev MLops and ML workflow platform for various computer-vision teams.

  • 2022
    Video-Enhancement Features
    • Webcam-denoising.
    • Artifact-management.

Amazonimageimage

2017-2021

I was employed as an Applied Scientist at Amazon Alexa AI. My focus was on efficient enterprise-level ML platforms that powered Alexa’s model-training infrastructure, used by over 1500 scientists with hundreds of models trained everyday. I primarily focused on cost-efficient, and scalable distributed training environments. I worked both in engineering and research capacities. The problem spaces that I focused on were neural network compression via network architecture search and predictive early-stopping algorithms. Some of the work I have done in this team have been featured at EMNLP 2020.

Previously, I was an applied scientist at Amazon Web Services AI Labs. I was part of the AWS SageMaker launch team and was involved in the development of several AWS Sagemaker CV algorithms, with primary ownership stake in Object Detection and Semantic Segmentation algorithms. I also was a member of the launch team of Sagemaker RL. I owned and launched model compression using RL that became a significant part of the Keynote address at re:Invent 2019. I also worked on domain-adaptation algorithms for Sagemaker CV, which lead to both products and publications including an oral paper at CVPR 2019.

For obvious reasons, some of the work that was performed while being employed by Amazon will not be made public here. Others that are already launched or made public are listed here.

Tenure Satistics:

  • Patents filed: 7
  • Peer-reviewed publications: 2
  • ArXiv papers/white papers: 1
  • Books/Book Chapters: 2

  • Jan, 2024
    Decoupled machine learning training.

    I have a patent Decoupled machine learning training on the cloud.

  • Nov, 2023
    Applying compression profiles across similar neural network architectures.

    I have a patent issued on applying neural network compression.

  • Sep, 2023
    Searching compression profiles for trained neural networks .

    I have a patent issued on searching for compressed neural network models.

  • Nov, 2022
    Reinforcement learning for training compression policies for machine learning models .

    I have a patent issued on neural network compression.

  • Oct, 2020
    Evaluating the effectiveness of ENAS for sentence-pair tasks.

    I have a paper accepted to the Workshop on Insights from Negative Results in NLP at the Conference on Empirical Methods in Natural Language Processing (EMNLP), 2020.

  • April, 2020
    Out-of-the-box channel pruned networks.

    I published an arXiv Pre-print titled "Out-of-the-box channel pruned networks".

  • March, 2020
    Domain mapping for privacy preservation.

    I have a patent issued on privacy preserving ML.

  • July, 2019
    d-SNE: Domain Adaptation using Stochastic Neighborhood Embedding.

    I published an oral paper at CVPR 2019. Code is available here.

  • Nov, 2018
    Amazon SageMaker Reinforcement Learning.

    I worked on the launch the Amazon SageMaker RL project for re:Invent 2018. One of the pieces that I worked on the launch of Neural Network Compression using Reinforcement Learning in collaboration with GE Healthcare.

  • Nov, 2018
    Amazon SageMaker Semantic Segmentation.

    I developed and launched the Amazon SageMaker Semantic Segmentation algorithms.

  • Aug, 2018
    Bring your own pre-trained MXNet or TensorFlow models into Amazon SageMaker.

    I published a blog demonstrating BYO Tensorflow or MXNet Models into SageMaker

  • July, 2018
    Amazon SageMaker Object Detection.

    I worked on the launch of the Amazon SageMaker Object Detection algorithms.

  • 2018
    Amazon SageMaker.

    In my duties with the Aamazon SageMaker team, I often contribute to @aws, @dmlc and @awslabs open-source repositories. I was part of the Sagemaker Launch team and contributed to several pieces of Software that were shipped at launch including the MxNet and tensorflow containers. Apart from simple issues, I have written several Sagemaker Example notebooks and have contributed to the Sagemaker python SDK.