profile-pic

Roshan Chaudhari

Roshan Chaudhari

Software professional worked on different layers of stack, from system level (graphics drivers) to application level. Expertise include OpenGL, Vulkan, GPU architecture, distributed computing, cloud microservices, autonomous driving, Lidars.

Currently working on making Nvidia cloud successful. Building an infrastructure for cloud microservices. Working on scaling cloud services.

  • Role

    Senior Software Architect

  • Years of Experience

    13 years

Skillsets

  • OpenGL - 6 Years
  • Distributed Systems - 4 Years
  • GPU - 5 Years
  • gRPC
  • TensorFlow
  • CMake
  • Protobuff
  • TensorRT
  • ONNX
  • Vulkan
  • GLSL
  • HLSL

Professional Summary

13Years
  • Oct, 2020 - Present4 yr 8 months

    Software Architect

    NVIDIA Graphics
  • Aug, 2018 - Sep, 20202 yr 1 month

    Senior Software Engineer

    DiDi Labs
  • Oct, 2015 - Aug, 20182 yr 10 months

    Senior Software Engineer

    Samsung Research America
  • Jun, 2011 - Jul, 20132 yr 1 month

    Software Engineer

    Imagination Technologies
  • Feb, 2015 - Oct, 2015 8 months

    Graphics Software Engineer

    Intel Corporation

Applications & Tools Known

  • icon-tool

    Docker

  • icon-tool

    Visual Studio Code

  • icon-tool

    Visual Studio

  • icon-tool

    GitLab

  • icon-tool

    Git

  • icon-tool

    gRPC

  • icon-tool

    Bazel

  • icon-tool

    ONNX

  • icon-tool

    Tensorflow

  • icon-tool

    TensorRT

  • icon-tool

    ROS

  • icon-tool

    OpenGL

Work History

13Years

Software Architect

NVIDIA Graphics
Oct, 2020 - Present4 yr 8 months

    o Building an infrastructure libraries for AI cloud microservices which has reduced development efforts by 50%.

    o Designed and architected cloud application, video interpreter, to allow video conferencing between people from different languages.

    o Designed and implemented gRPC async server, client and gstreamer libs which are leveraged by different teams in NVIDIA.

    o Increased scalability by 200% by finding out bottlenecks to support hundreds of concurrent sessions. This has reduced deployment costs by 50%.

Senior Software Engineer

DiDi Labs
Aug, 2018 - Sep, 20202 yr 1 month

    As a founding member of the infrastructure team for the autonomous driving stack, built an infrastructure supporting all the modules in the stack which has reduced development efforts by 30%.

    Designed and implemented sensor modules to process large amounts of data in latency sensitive manner.

    Improved inference time for machine learning models by converting them to TensorRT from Tensorflow, this included contributing to open-source projects for ONNX parsers.

Senior Software Engineer

Samsung Research America
Oct, 2015 - Aug, 20182 yr 10 months

    Improved performance of the engine by profiling and finding bottlenecks in the framework.

    Implemented batching of draw calls to improve the performance by 30%.

    Implemented foveated rendering with multiple layers to improve performance by 20%.

    Technology Summary: C++, OpenGL, Vulkan, OpenGL-ES, GLSL, HLSL

Graphics Software Engineer

Intel Corporation
Feb, 2015 - Oct, 2015 8 months
    Implemented test suit and framework for testing kernel mode graphics drivers. This also included debugging drivers for various issues.

Software Engineer

Imagination Technologies
Jun, 2011 - Jul, 20132 yr 1 month
    Developed applications to test embedded systems drivers using OGL, EGL, OpenGL-ES and GPU shading language features.

Major Projects

4Projects

Maxine Cloud

Oct, 2020 - Present4 yr 8 months
    Building an infrastructure for AI cloud media microservices and applications. Built a cloud application using multiple microservices for real-time conference use-case. Implemented gRPC async server, client, and other supporting libs for the microservices which are used across NVIDIA. Increased scalability by 200% by finding out bottlenecks to support hundreds of concurrent sessions. This saved AWS infrastructure costs significantly.

Deep Learning Super Sampling Framework (DLSS)

Oct, 2020 - Present4 yr 8 months
    Implemented translation libraries to enable Deep Learning SuperSampling (DLSS) with DX12 on Proton. This included contributions to open-source libraries for NvAPI and vkd3d-proton as well as NVIDIA proprietary drivers.

Autonomous Driving Stack Infrastructure

Aug, 2018 - Sep, 20202 yr 1 month
    Built an infrastructure supporting all modules in the autonomous driving stack. Improved inference time for machine learning models by converting them to TensorRT from Tensorflow.

Foveated Rendering

Oct, 2015 - Aug, 20182 yr 10 months
    Implemented foveated rendering with multiple layers to improve performance by 20%.

Education

  • Masters in Computer Science

    Arizona State University, USA (2014)

  • Bachelors in Information Technology

    College Of Engineering, Pune (2011)

  • Masters in Computer Science

    Arizona State Univesity (2014)

  • Masters in Computer Science

    Arizona State University (2014)

  • Bachelors in Information Technology

    College Of Engineering Pune (2011)

Interests

  • Cricket
  • Travelling