Work Experiences
- Under the supervision of Prof. Kenji Shimada.
- Synthetic point cloud data generation using CAD/OBJ models.
- Developing machine learning algorithms for Point Cloud data.
- Object detection and warpage measuring using PCL.
- 3-D reconstruction using Multi-light images for surface inspection.
- Machine Learning & Artificial Intelligence for Engineers (24-787)
- Preparing and teaching recitations.
- Holding office hours to assist students clarifying doubts.
- Grading and preparation of quizzes, problem sets and other assignments.
- Mentor groups of students with their course project.
- Assisting instructors with other activities associated with teaching a course.
- Maintaining website for the course.
- Intro to Scientific Computing (24-281)
- Assist professor with Homework, Quizzes, Projects and Tests.
- Communicated problems and updates with fellow course assistants and professor.
- Hand-on experience with soft material fabrication.
- Designed robotic hand gripper and an ultra-sensitive tactile sensor.
- Soft gripper could lift 200 times its own weight and sensor was sensitive to 0.5 mN force.
- Hands-on experience with ROS, C++, and Linux.
- Programmed Arduino to control mini robots.
- Developed automatic wireless communication between micro and mini robot.
Projects
- Deep Learning based approach to translate scanned documents/text in wild images from A to B language.
- Developing a complete pipeline to perform OCR on images and perform word-by-word translation.
- LSTM based Sequence-to-Sequence Deep learning model for translation.
- Aims to improve translation results on erroneous OCR detections.
- Investigating improvements in SLAM algorithm by detecting and removing Dynamic objects.
- Verifying results on KITTI and TUM-RGBD dataset and ORB-SLAM2 and LSD-SLAM.
- Deploying video-based inpainting Deep Learning models.
- Developed a pipeline for a large scale data set (~1Tb).
- Algorithms based on Linear Regression, Decision Tree, Random Forest and XGBoost.
- Using PySpark, Amazon AWS and Microsoft Azure, able to achieve an RMSE of 0.73 $.
- Detailed comparison study between different models based on model size, RMSE, inference time etc.
- Developing a 2-step authentication model to learn and verify the user based on the typing patterns.
- Algorithms based on SVM, Neural Networks, Decision Trees.
- Implementing and developing algorithm to boost content aware image resizing
- Compression of the image size by upto 70%
- De-noised the input sensor data using Kalman Filter.
- Developed PID, Feedback, Optimal controller for the vehicle and bagged position in top 20 %
- Developed a tiny package to take on screen inputs without using external library.
- Integrated both numerical and graphical problem solving.
- Invented a prototype to show working of a novel real-time spinal adaptive smart bed.
- Integrated Inertial Measurement sensors, Infrared sensor, wireless control, Arduino.
- Developed a line follower and obstacle sensing robot.
- Operable through Bluetooth and used Arduino and wireless controlling using XBee.
- Analyzed the mechanics and electronics integration.
- Integrated with proximity sensors, gyroscope, and wireless control.