Kunal Agarwal

Robotics & Computer Vision enthusiast

Hi, I am Kunal Agarwal
I have completed my B.Tech in Computer Engineering from Veermata Jijabai Technological Institute, Mumbai, India.
I am currently working as Software Engineer at VISA

My Research Interests are:

  • Robotics and Automation: SLAM (Simultaneous Localization and Mapping), Autonomous Vehicles
  • Perception and Path planning for Autonomous Systems

You can reach me out on Matrix @kunalagarwal18:matrix.org or email me at kunalagarwal1072002@gmail.com

Experience

VISA Inc, India (Data & AI Platform)

Software Engineer | prev Summer Intern
June 2024 - Present | June 2023 - Aug 2023
  • Automated the Release Review process within CI/CD frameworks, a solution now adopted by 70% of the company’s teams, reducing workload and saving three days of manual effort per release review for both Release Management and Development teams.(GitHub Actions, Jenkins, Javascript, Groovy)
  • Working on fullstack for Creating a LLM-based ATC Buddy, a 24/7 smart assistant for Application Technical Contacts (ATCs) that streamlines application ownership tasks and reduces lead times, useful across all VISA departments. (LLMs, Langgraph, python, React)

Multi-Robot Autonomy Lab, IISER Bhopal

Robotics Research Intern
Jan 2023 – Dec 2023
  • Worked on Project - 'CFDMPC: MPC-based UAV Path Planning Algorithm With CFD-Based Wind Field Estimation'
  • Designed a Model Predictive Controller (MPC) using the Multiple Shooting Method to move the UAV along an optimized trajectory against a constant wind flow and implemented obstacle avoidance using the MPC controller
  • Developed a Physics-Informed Neural Network(PINN) model using DeepXDE to solve steady-stade Reynolds-averaged Navier Stokes(RANS) equation, for generating flow field in the UAV workspace with varied wind patterns around obstacles
  • Drafted a research paper on it, preparing for submission to a journal

Libcamera @ Google Summer of Code 2022

Open Source Developer
June 2022 – Nov 2022
  • Implemented GPU-based software ISP(Image Signal Processor) within libcamera framework to make it compatible with devices that doesn’t have a hardware ISP.
  • Used Generic Buffer Management(GBM) utility of Mesa to export buffers for GPU to be used as framebuffers for rendering textures including the use of dmabufs to achieve zero-copy of data in memory
  • In the Software ISP, All Image processing algorithms like Demosaicing, Auto-white balance, Auto-contrast and so on are implemented using OpenGL Compute Shaders to perform all the computations in GPU
  • Report: summerofcode.withgoogle.com