Download Resume:  PDF.

AI Engineer
Reality AI, Columbia – MD

Email: kajaldamjigada-at-gmail-dot-com


    Libraries: Robot Operating System (ROS) • OpenCV • NumPy
    Simulation Environment: V-Rep • Gazebo • RViz
    Tools: Git • Microsoft Visual Studio • Eclipse
    Operating System: Windows • Linux
    Over 5000 lines: Python • C++ • Matlab • LaTeX
    Familiar: C • Java


  • Master of Engineering, Robotics – May 2017
    University of Maryland, College Park, MD
  • Master of Business Administration, Technology Management – May 2015
    NMIMS University, Mukesh Patel School of Technology Management & Engineering
  • Bachelor of Technology, Electronics and Telecommunication – May 2014
    NMIMS University, Mukesh Patel School of Technology Management & Engineering


  • Qualcomm Technologies Inc. – Engineer (January 2018 – present)
    • Developed applications as software module in C++ and implemented on aerial and ground robots using ROS.
    • Built software module for Qualcomm chip 8×96 using docker for cross-compilation.
    • Debugged hardware interfaces and real-time protocols between software modules for smooth application runs.


  • Reality Analytics, Inc. – AI Engineer (June 2017 – December 2017) & AI Intern (Jan – May’ 17)
    • Applied machine learning techniques for object detection and classification in images and videos using MATLAB.
    • Developed a function to extract various objects from images to create a database of templates.
    • Machine Learning techniques – normalized cross-correlation, SVM and neural network.


  • Autonomy, Robotics, and Cognition Laboratory – Graduate Assistant, University of Maryland (Jan – May’ 16)
    • Experimentally evaluated path planning algorithms – RRT and RRT* with Open Motion Planning Library (OMPL).
    • Tested the algorithm to move Baxter’s arm with collision avoidance using motion planning framework Move-It on ROS.

PROJECT – Vision and Perception:

  • Texture Synthesis by non-parametric sampling (May 2016)
    • Created a large image of a given texture by placing close match pixel neighbours around the sample over multiple iterations.
    • Tested using multiple texture samples for varying window size and threshold values.
  • Panorama image by image stitching using SIFT features (March 2016)
    • Identified the matching features (SIFT) between images to obtain the affine transform that aligns the adjecent images.
    • Evaluated the best affine transform using RANSAC and used it to stitch the images for creating a panorama.
    • Enhanced the final image to a single intensity using normalization and filled up the empty spaces using bi-linear interpolation.

PROJECT – Path Planning and Navigation:

  • Localizing a robot using particle filter approach with Python (April 2016)
    • Implemented particle filter on the laser scan data for estimating the robot’s position and orientation in the map.
    • Reduced the sensing noise by comparing the actual vs expected laser scan data, computed using robot’s odometer reading.
  • Map building from laser scan data and vizualized in Rviz (April 2016)
    • Implemented an occupancy grid-based approach to compute the map of the environment for a mobile robot.
    • Identified the occupied (or obstacles) cells on the map by combining the laser scan data with the robot’s odometry.
    • Used RViz in Robot Operating System (ROS) to visualize the mobile robot and the mapped environment.
  • Path Planning and Control for Turtle-bot using ROS (March 2016)
    • Developed an algorithm to compute path for the robot and designed a proportional controller to drive the robot.
    • Implemented the algorithms in ROS using Python and tested its performance in Gazebo simulation environment.
    • Evaluated the performance of the developed algorithms by implementing it on a physical Turtle-bot.

PROJECT – Machine Learning:

  • Teaching a computer to play Tic-Tac-Toe with Q-learning and Neural Network (December 2016)
    • Programmed to learn the winning strategy of Tic-Tac-Toe by playing the game with itself on a reward-based methodology.
    • Compared the result with win-loss based method of neural network and found both to be effective.
  • Learning a function based on input-output point using CMAC (September 2016)
    • Implemented an Cerebellar Model Arithmetic Computer (CMAC) method to approximate a function using.
    • Designed based on input and output data points over various iterations to drive error to zero.