Saurabh Palande

I’m a Robotics Software Engineer specializing in computer vision and GPU-accelerated perception systems. My work focuses on building and deploying real-time vision pipelines for industrial systems, where performance and reliability are critical particularly in applications like aircraft sanding. I have hands-on experience developing custom CUDA kernels, optimizing GPU workloads using profiling tools like Nsight Compute, and designing perception systems that integrate RGB-D cameras, LiDAR, and embedded platforms such as NVIDIA Jetson. My background spans mapping, localization, and sensor fusion, with a strong emphasis on translating algorithms into efficient, real-world systems. I’m especially interested in solving challenging problems at the intersection of robotics, perception, and high-performance computing.

Education

  • Master of Engineering in Robotics

    University of Maryland, College Park

    May 2023

  • Bachelor of Technology in Mechanical Engineering

    VNIT, India

    May 2019

Work Experience

Robotics Software Engineer (Computer Vision)

Temple Allen Industries, Rockville, MD

June 2023 - Present

  • Implemented custom CUDA kernel functions for a smart automated pneumatic sanding system used in aircraft hangars to significantly boost computer vision pipeline performance, achieving substantial improvements in real-time processing and operational efficiency critical to precision and safety in aircraft sanding operations.
  • Utilized GPU profiling tools (Nsight Compute) to analyze and optimize CUDA kernels, identifying memory-bound vs compute-bound kernels, bottlenecks, and improving overall performance.
  • Implemented and optimized custom algorithms for the Robot Body Filter and Proximity Detection System, enhancing the robot’s ability to isolate itself from camera views and navigate safely. Worked with NVIDIA Jetson platforms for deploying optimized perception algorithms in embedded systems.
  • Investigated and worked with various RGB-D cameras and depth sensing technologies (Stereo, IR, ToF), and LiDAR systems, evaluating and selecting optimal sensors and designing camera/LiDAR placement to ensure comprehensive environmental coverage and system safety.
  • Developed CUDA-based mapping algorithms to process and denoise point cloud data, enabling robust environment perception and improving detection reliability.

Robotics Engineer Intern

Kick Robotics, College Park, MD

September 2022 - May 2023

  • Developed a python script for parsing raw GPS data and converting it to ROS2 message format.
  • Refined state estimation of the robot by fusing GPS data with IMU and odometry and improved loop-closure accuracy using the robot localization package in ROS2. Successfully deployed and tested the same on a real robot.

Robotics/Machine Learning Intern

Soterya, Inc, Boston, MA

June 2022 - Aug 2022

  • Developed a smart robotic mattress to decrease the risk of SUDEP-major cause of death in young epilepsy patients.
  • Implemented Human pose detection algorithm using Machine Learning to detect the prone position from the heatmap generated from the pressure sensors to reposition a patient into a recovery position.

Research Assistant

Gamma Lab - University of Maryland, College Park, MD

Feb 2022 - May 2022

  • Built packages in ROS2 for the localization (sensor fusion and state estimation), motion planning, and navigation of an autonomous indoor delivery robot while working as a Research Assistant in Prof. Dinesh Manocha’s lab.
  • Created training environments to simulate intelligent pedestrians in Gazebo to match the real-world scenario in a crowded indoor place. Designed CAD models in SolidWorks and 3D printed the parts for the robot.

Computer Vision Intern

Shyena Tech Yarns, Pune, India

March 2021 - June 2021

  • Researched and implemented deep learning models used for Human Activity Recognition (HAR) and 3D object classification, segmentation, and detection on point cloud data generated by LiDar sensor.
  • Developed an automated machine learning pipeline for 3D object classification and segmentation.