• Designed and owned core controls and goal planning for autonomous shipping pallets running ROS2 C++ on Nvidia Jetsons.

• Built hierarchical state machine to resolve concurrency issues regarding asynchronous sensor and actuator data streams

• Introduced test-driven development practices and CI pipelines using Gtest/Gmock and GitHub Actions to improve reliability of robotics software stack

• Developed firmware for battery management, sensor filtering, and live firmware updating for STM32 MCUs

• Integrated multiple VSLAM sources with motor feedback using Kalman filtering for accurate odometry estimation

• Maintained Perl system for ingesting job listings from various partner systems into AWS data systems.

• Re-designed the metrics API using Go and Scala to achieve 100X increase in batch call limit within the same return time.

• Automated emails for metrics reporting to clients via Python.

• Developed legal compliance service for reporting and removing user data.

• Set up scripts for migrating the onboard computer of a hexamotor autonomous drone from Snapdragon Flight Pro to ModalAI VOXL.

• Developed infrastructure tools enabling for compilation and execution of ROS nodes directly on the drone or within Docker emulators.

• Implemented vision algorithms for camera sensor with single core SoC using OpenCV and C++.

• Created example lane following project for miniature Arduino car with PID controller.

• Created basic object detection, tracking, and prediction algorithm via blob detection and/or key-point matching combined with Kalman filtering.

• Implemented AprilTag fiducial marker recognition.

Systems integration on autonomous car platform involving toy cars

• Updated software packages to operate miniature autonomous car platform with sensors.

• Used 2D laser scanner to construct map of obstacles around vehicle.

• Integrated package for constructing a map with probabilities of obstacles given noisy sensor data.

• Created a package that can generate geometric paths on a map, created from Lissajous curves.

• Implemented feature that prevents vehicle from deviating from a given path during certain edge cases.

• Set up 3D laser scanner with localization and mapping packages for vehicle.