Vishesh Jain

Developed a YOLOv5-based object detection pipeline in ROS2 for autonomous robotic perception, achieving sub-100ms inference latency. Built sensor fusion modules combining camera and IMU streams using SLAM, enabling accurate obstacle detection and autonomous navigation.

AI/ML & Robotics Intern at Mafkin Robotics Pvt. Ltd. (2026-01 – 2026-04)

• Developed a YOLOv5-based object detection pipeline in ROS2 for autonomous robotic perception, achieving sub-100ms inference latency.
• Built sensor fusion modules combining camera and IMU streams using SLAM, enabling accurate obstacle detection and autonomous navigation.

Deep Learning Intern at NIT Patna (2025-05 – 2025-07)

• Trained a ResNet-based image classifier with transfer learning and channel attention on 10,000+ samples (CCMT dataset), reaching 91.14% top-1 accuracy across 38 disease classes.
• Built an augmentation and z-score normalization pipeline to curb overfitting; tuned AdamW with cosine LR scheduling and deployed the model into an AI-powered Smart Irrigation System.

B.Tech in Computer Science – Vellore Institute of Technology (VIT), Chennai (2023-08 – 2027-08)