• Worked on the Behavior Planner and Constraints Generation for highway features like merging, split-handling, tactical and mandatory lane changes.

• Developed and deployed an LLM-based scene reasoning module that translated raw vehicle signals and image data into ranked root-cause analyses with explanatory narratives, improving the accuracy of traditional rule-based methods by 9 percent.

• Applied supervised and unsupervised Machine Learning techniques on large-scale fleet data to predict root causes of BlueCruise system cancellation events, driving improved system reliability.

• Designed and implemented an anomaly detection pipeline to identify novel, safety-critical events from production vehicle logs.

• Re-factored the Behavior Selection module to handle stay-in-lane and lane change behaviors.

L4 Autonomous Shuttle:

• Created a remote route-request/take-over pipeline as part of the Autonomous Shuttle project to allow remote teleoperation of the AV shuttle

• Enhanced the reliability of planning and decision-making algorithms to effectively handle intersections and start/end of route scenarios for the AV shuttle project.

• Improved route and motion planner robustness by adding replanning logic to handle dynamic and uncertain environments.

• Devised and demonstrated a business case prototype for autonomous parking of fully assembled vehicles in manufacturing plants.

• Implemented a controlled safety stop-trajectory of the Autonomous vehicle when encountering specific failure criteria.

• Led research collaborations with universities on aspects of Motion Planning

C++ | Python | ROS | AOS | Jira | Agile | Carla Simulation

• Collaborated on the development of a C++ Trajectory Planner and Controller library, which encompassed lateral features like Lane Keeping and Lateral Collision Avoidance. This involved integrating environmental signals, driver inputs, and vehicle states.

• Constructed a C++ pipeline for Ethernet-based UDP communication on the dSpace Embedded PC, enabling real-time execution of C++ executables. This setup facilitated seamless interaction with the dSpace MicroAutoBox for Vehicle I/O data via CAN communication.

• Utilized CAN tools (CANoe, CANalyzer, CANdb++, etc.) to capture and analyze CAN data for the integration of features such as Adaptive Cruise Control (ACC) and Traffic Jam Assist (TJA) into test vehicles.

• Assisted in the development of the State Machine responsible for feature arbitration between various longitudinal and lateral functions, including ACC, Autonomous Emergency Braking (AEB), and Lane Centering.

• Created CarMaker simulation environments to validate the performance of features such as ACC and TJA in diverse traffic and road scenarios.

• Developing control strategies for path-tracking and deploying them on test vehicles using dSPACE solutions for prototyping and implementation.

• Conducting research on planning algorithms for advanced driver assistance and semi-automated driving features like Adaptive Cruise Control and Traffic Jam Assist.

• Using CAN tools (CANoe, CANalyzer, CANdb++, etc) to record, analyze and, manipulate CAN data for integration of features in test vehicles.

• Using CarMaker simulation environment to validate the performance of Automated Driving features.

• Using GIT and PTC for Version Control (Keeping track of model development, code update, and documentation).

• Working on automation scripts for vehicle parameter characterization.

• Built a simulation script to evaluate the Sensor Fusion algorithm by associating the Radar and Camera object data frames.

• Worked on the Target Object Selection Algorithm based on the Ego Vehicle Path Prediction.