Project: E-Scooter Trajectory Prediction Study•Created LiDAR-Radar-Camera sensor processing pipeline for speeding up data annotation tasks by 4.5x.•Optimized 110 hours of driving data from 4 cities across the US resulting in increase of quality training data and cost reduction of over $100k.Project: Alcohol-Impaired Driving Study•Developed a driving data collection suite utilizing Robot Operating System (ROS) and a custom GUI on a Toyota Avalon for an Alcohol-Impaired Driving Study analyzing 15 human subjects driving in real-world scenarios under alcohol influence.•Implemented a real-time data capturing system from sources including multiple cameras, GPS, vehicle CAN bus, microphone, and alcohol breath sensor capturing 5TB+ of synchronized data while minimizing manual user intervention.•Constructed algorithms to maintain 98% data synchronization from multiple sensors and polling intervals.Project: C-V2X Traffic Management Study•Spearheaded a team of 3-research assistants, collaborating with the Indiana Department of Transportation (INDOT) to create a Cellular Vehicle-to-Everything(C-V2X) prototype for enhancing flow of traffic information to connected vehicles(CV).•Designed geo-fencing algorithms utilizing RabbitMQ and MQTT to enable targeted routing of traffic data and reducing message congestion in vehicular networks by 30%.•Conducted track testing on vehicles to verify feasibility of C-V2X in real-world usecases utilizing custom made metrics monitoring system.

• Streamlined RESTful APIs for backend services using Spring Boot and Swagger to ensure consistency among multiple cloud applications in the SAP Enterprise Product Development (EPD) suite.• Formulated asynchronous multi-processing techniques to speed up a document retrieval backend service by a factor of 5x.• Utilized platform-agnostic frameworks and domain design principles to engineer cloud-native apps and reduce deployment effort when releasing on multiple cloud IaaS providers (AWS, GCP). • Restructured application deployment with CI/CD pipelines (Jenkins) and ensured documentation and test coverage by over 90%, resulting in a 40% reduction in deployment time for new features.• Optimized pre-made ML models using SAP AI Core, cutting down model training time by 15% while keeping inference response time under 1s.

• Devised backend REST APIs using Python-Flask for consumption of Named-Entity-Recognition (NER) microservice running on the SAP Cloud Platform.• Decreased inference response time by over 50% by developing a custom performance metrics monitoring module resulting in a 10% cost reduction for consumers.• Modeled HTTP requests utilizing Swagger API to maintain consistency and clarity in API design.

Designed and Developed a Court Reservation System for indoor sports courts utilizing a Full stack solution(LAMP).