▪Worked on generation of camera trajectory data for rendering 3D graphics in the camera scene using structure from motion▪Worked on estimating the ball trajectory for a ball hit for six or catch in a cricket match using broadcast stream▪Worked on CNN based approach for converting a 16:9 video stream to 9:16 video stream focusing on salient region

▪ Implemented video synchronizer, that match actions in two videos▪ Trained ML model for player detection in NBA team using transfer learning on AWS▪ Converted PyTorch model to CoreML model with pre and post-processing using coremltools library▪ Build an iOS swift package for object tracker module▪ Experience with Jira, Confluence, and GitHub

▪ Develop and maintain structure from motion algorithm for generating a 3D model using forward-looking and 360 videos on Linux machine▪ Work with point cloud processing libraries VTK, PCL, open3d to align, filter, segment, and register point clouds▪ Train and modify CNN using TensorFlow, TensorFlow Lite, Keras for desktop and mobile application▪ Worked on developing a mobile device application for the generation of the 3D model in real-time on Android and IOS▪ Worked with Mayo… Show more ▪ Develop and maintain structure from motion algorithm for generating a 3D model using forward-looking and 360 videos on Linux machine▪ Work with point cloud processing libraries VTK, PCL, open3d to align, filter, segment, and register point clouds▪ Train and modify CNN using TensorFlow, TensorFlow Lite, Keras for desktop and mobile application▪ Worked on developing a mobile device application for the generation of the 3D model in real-time on Android and IOS▪ Worked with Mayo Clinic on 3D CNN for lesion detection▪ Worked on developing algorithms for modeling a custom ear-bud which include generation of 3d ear model, ear and ear landmark detection using neural network, segment region of interest, registration of generated model to database ear scanned models▪Experienced in maintaining version control using GitLab Show less

▪ Developed an image segmentation algorithm to separate foreground and background▪ Studied different feature extraction algorithm (time complexity) like SIFT, SURF, ORB, KAZE, and AKAZE to create an unique id for feature matching▪ Designed a User Interface using Tkinter along with multithreading▪ Trained a neural network for hand detection with TensorFlow Object Detection API using a pre-trained model

Subject: Applied Optics and Photonics

Studied about the working and maintenance of entertainment system in an aircraft.