Pittsburgh, Pennsylvania, United States
Undergraduate student at Carnegie Mellon University studying Computer Science interested in robotics and software development.
- Managed 50+ computer engineering, computer science and AI students to develop an autonomous racecar - Pioneered driverless in North America by helping FSAE introduce first NA driverless competition and promote the growth of other teams by providing a development framework for autonomous cars - Led documentation for entire software stack and open-source release viewable at cmr.red/driverless-docs - Made design decisions for mechanical, electrical and software systems (sensor selection, algorithm design, software interfaces) - Productionized software stack by introducing caching, shared memory parallelism and GPU partitioning
- Incoming Vice President of Driverless, CMR (June 2024 onwards) - Led new members to create testing framework for ROS2 nodes to ensure edge-case performance of software components like data association, Delaunay triangulation and raceline generation - Researched and developed SLAM pipeline based on iSAM2 algorithm [ROS2 / C++] - Leveraged Support Vector Machines (SVMs) to generate the midline between cones of different colors in real time - Translated EKF-SLAM and raceline generation pipeline from Python to C++ using Eigen, GSL2
- Designed Software-in-the-Loop, debugging / unit-testing environment for vehicle controls algorithms - Implemented Hardware-in-the-Loop, similar testing environment for custom PCBs controlling the powertrain, driver interface, etc. by simulating digital I/O and CAN communication on a microcontroller [FreeRTOS]
Lead weekly labs, recitations, office hours and debug student code for Parallel and Sequential Data Structures & Algorithms which covers recurrences, trees, randomization, shortest paths, contraction and dynamic programming
Lead weekly labs, recitations, office hours and debug student code for Principles of Imperative Computation, a data structures and algorithms class in C
- Designed software for sensor-equipped robot arm to (a) follow surface defects by generating and analyzing point cloud information in real time, (b) conformally 3D-print by adding a 2-axis rotating platform - Generated gaits for modular robot (legs, wheels; up to 18 DOF) using RL algorithm proximal policy optimization; trained on thousands of robot instances with NVIDIA Isaac Gym simulation inside Docker container - Developed Python config manager to address automatic directory creation, YAML discovery and variable scoping