Advanced Automotive AI System Development:• Implemented a comprehensive perception system for Honda's autonomous vehicle prototype, integrating multiple deep learning models for object detection, segmentation, occlusion, warehouse locating and tracking.• Developed a continual learning framework for ADAS object detection models, enabling adaptation to new road conditions and vehicle types without compromising performance on existing tasks.• Implemented self-supervised learning techniques to leverage large amounts of unlabeled driving data, significantly improving model robustness in diverse weather and lighting conditions.• Created a diffusion-based system for synthetic data generation, addressing class imbalance issues in rare traffic scenarios and enhancing overall model performance for autonomous driving.Applied Automotive Projects:• Led the development of an end-to-end quality control system for Honda's vehicle manufacturing line, combining defect detection, panel gap measurement, and paint quality inspection models.• Designed and implemented a real-time multi-sensor fusion system for 3D object detection in Honda's ADAS, integrating data from cameras, LiDAR, and radar sensors.• Developed an advanced driver monitoring system using multi-task learning, simultaneously performing gaze estimation, emotion recognition, and drowsiness detection.• Implemented a visual SLAM (Simultaneous Localization and Mapping) system for precise vehicle localization in GPS-denied environments.AI Integration and Deployment:• Collaborated with Honda's vehicle electronics team to optimize and deploy AI models on automotive-grade hardware, ensuring real-time performance and compliance with ISO 26262 safety standards.• Implemented federated learning techniques to improve AI models using data from Honda's fleet while preserving customer privacy.

• Developed machine learning models for predictive maintenance of Honda vehicles, reducing unscheduled downtime by 12%.• Implemented a graph-based recommendation system for Honda's online vehicle configurator, increasing customer engagement and option uptake by 17%.t target value.

• Forecasted price and required volume of raw material, finishing date, and expected outstanding• Made integrate report on design, fabricate, erection and purchase schedule, coordinated with other sub-contractors to update and adjust upon overall schedule.