Nithin Shivanna

Project: Object Type Classification using Radar Data

• Designed a radar-based object classification system using advanced deep learning techniques to interpret radar detected object location and reflection data for high-accuracy classification.
• Executed full lifecycle of AI model deployment on embedded systems, from data preparation to post-training parameter extraction.
• Developed and deployed neural network algorithms in C++ to be integrated into embedded ADAS platforms with high real-time performance.
• Built visualization tools for performance analysis and automated model parameter exporting, ensuring seamless integration into target software.
• Worked on an earlier-generation Naïve Bayes classifier, contributing to radar-based object classification improvements.

Project: Scene Type Classification using Radar Data

• Designed and implemented a Graph Neural Network (GNN) architecture tailored for radar-based scene classification.
• Automated the end-to-end pipeline for continuous model training, validation, and evaluation, reducing manual intervention and accelerating experimentation.
• Developed custom tooling for radar data annotation, leading to significant improvements in dataset quality and model performance.

Project: Radar Sensor Fusion (Radar + Video)

• Participated in building sensor fusion systems combining radar and camera inputs, using Kalman Filters and association logic to improve tracking and decision accuracy.
• Calibrated gating parameters, improved fusion accuracy, and validated results on highways and test tracks.
• Synced multi-sensor data to ensure effective sensor alignment and timing correctness.

Project: Simulation Tool Development – Radar Applications.

• Developed a simulation toolchain for radar applications, enabling virtual testing and debugging of ADAS functionalities without hardware dependency, and improving efficiency in the development cycle.
• Designed and implemented evaluation tools to analyze simulation results, providing KPI metrics for performance assessment and optimization.

Project: Cisco Immersive Telepresence Endpoints

• Worked on the development of the Call Control Agent, UI Manager, Web Server, and Media modules, contributing to core feature enhancements.
• Led the development of application features such as Proximity, Configurable Presentation Stream, Ad hoc Conference, and TMS Phonebook.
• Developed automation scripts using Python and the Robot Framework to streamline testing and validation.

Project: NSN Flexi Lite BTS

• Worked on regression testing to ensure software stability and performance in outdoor micro-cell environments.
• Identified and fixed defects, improving reliability for 3G and LTE (FDD/TDD) deployments.