At Swaja Robotics, I led development of a real-time PCB defect inspection pipeline using YOLOv8, covering data annotation, model training, benchmarking, and deployment. I also developed an audio anomaly detection system with transformer-based architectures and deployed it on embedded hardware.
Earlier, as a Machine Learning Intern, I developed a glass-break detection system and deployed it on Nordic Thingy:53 using TensorFlow Lite and Edge Impulse. I also built a student guidance recommendation system that used ensemble models to generate personalized academic and career reports.
