Abhinav Grover

Abhinav Grover

MASc Student and Researcher

University of Toronto Robotics Insitute


I am Machine Learning Engineer at Kindred AI working to build the wolrd’s first autonomous system for grocery fulfillment. In the past, I was a research student at University of Toronto, adviced by Dr. Jonathan Kelly where I had the opportunity to publish several papers in the field of Robotics and Artificial Intelligence. I am a receipient of the NSERC Canadian Graduate Scholarship and Vector Scholarship in AI.

I graduated from University of Waterloo with a bachelors in Mechatronics Engineering and an option in AI. I worked as a research assitant for Dr. Krzysztof Czarnecki, Dr. Mihaela Vlasea, and Dr. James Tung. For 12 months, I worked as an intern at Nvidia US where I contributed to their self-driving infrastructure. Through various co-ops and internships, I graduated with over 2 years of work experience.

Please checkout my projects and publications below.

CV \ résume.

  • Autonomous Robots
  • Robot Manipulation
  • Artificial Intelligence
  • Cricket, Tennis and Badminton
  • MASc in Aerospace Engineering (Robotics), 2021

    University of Toronto

  • BASc in Mechatronics Engineering, 2019

    University of Waterloo


Machine Learning Engineer
Sep 2021 – Present San Francisco, California
We are building the world’s first grocery fulfillment robotic system capable of being fully autonomous.
Graduate Research Student - Robotics and AI
Sep 2019 – Aug 2021 Toronto, Canada
I worked as a researcher at STARS lab under the supervision of Dr. Jonathan Kelly. For my project, I created a learning-based algorithm to perceive in-hand object slip using inexpensive barometric tactile sensors. A publication on this topic, submitted to ICRA 2022, is currently under review. During the course of my masters, I have received the NSERC Canadian Graduate Scholarship and the Vector Scholarship in AI.
Software Engineer - Autonomous Driving
Jan 2018 – Aug 2018 Holmdel, NJ

Worked in a team of 25 machine learning engineers, led by Dr. Urs Muller, creating an end-to-end autonomous driving solution. (Linux, C++, Git, Bash)

  • Contributed to the re-architecture of Nvidia’s Driving Simulation application in order to enable simulation on the GPU cluster and standardize the benchmark testing of DriveNets across the company
  • Developed a prototype of a model-predictive vehicle controller in C++ that runs on the DRIVE Xavier platform
  • Contributed to Nvidia DriveWorks SDK which meant overcoming extensive quality checks designed for MISRA compliance
Systems Engineer - DRIVE platform
May 2017 – Aug 2017 Seattle, WA

Worked on Nvidia’s DRIVE hardware stack, aiming to revolutionize the Automotive computing industry (C++, bash, python)

  • Contributed to the Embedded PDK for the DRIVE automotive hardware products, that now serve giants like Tesla and Audi.
  • Re-implemented the Tegra flashing software for the DRIVE computing platforms to meet strict Automotive standards and to gain MISRA compliance.
  • Architected the incremental Tegra flashing functionality, reducing the average OS flash time to half.
Software Engineer - Special Projects
May 2016 – Aug 2016 Waterloo, Canada

As a core member of the software development team, I worked closely with the R&D department to prototype a new application for their tool tracking system.

  • Prototyped a cranial navigation Mac OS application using the existing tracking system
  • Worked with OpenGL based Visualization Toolkit (VTK) for 3D image rendering
  • Gained understanding of computer vision algorithms used for sphere tracking
  • Exposed to optimisation algorithms used for shape matching


Accurate Road Segmentation using Camera and LIDAR Data for self-driving application

Accurate Road Segmentation using Camera and LIDAR Data for self-driving application

Completed as part of a graduate course at the University of Toronto | PyTorch. Implemented a Fully Connected Network based Road Segmentation pipeline on Audi’s A2D2 dataset using PyTorch and OpenCV.

Failure-Mode Analysis of A Learned Dexterous Hand Controller

Failure-Mode Analysis of A Learned Dexterous Hand Controller

Completed as part of a graduate course at the University of Toronto | Tensorflow, Python, OpenAI gym. Conducted a failure mode analysis on learned DDPG-based policies used to control a dexterous hand with tactile sensors, in an attempt to understand the utility of tactile information.

Invariant EKF Slam

Invariant EKF Slam

Completed as part of a graduate course at the University of Toronto | MATLAB. Simultaneous Localization and Mapping (SLAM) has been a highly research problem and the techniques that solve it have undergone huge improvements in the last two decades.