Mridula Vijendran

• Using LLMs and RAGs for reliable document extraction, web scraping, automation of tasks.
• Working with multimodal systems and chatbot design for reliable, accurate responses.
• Streamlined data preprocessing pipelines by automating routine tasks with Python scripting, reducing time spent on manual processes significantly.

I was the teaching assistant for below units:

• High Performance Computing
• Robotics
• Computational Thinking
• AI and Data Science

• Self-supervised 3D Keypoint learning for egomotion estimation: paper reproduction. For visual odometry using pytorch. Involves building the keypoint net and depth net models in a pipeline with multiple losses and preprocessing for feature tracking and keypoint warping for 3D lift to measure depth of scene.
• Pytorch projects on image segmentation and optical flow.

• Communicated effectively with faculty and staff and accepted critiques and suggestions for areas of improvement.
• “Applied Deep Learning for Image Analysis” – Effectiveness of a Convolution Neural Network (CNN) for detecting harmful objects in the images
• Keras (Python Deep Learning Framework), Web Scraping, Dataset cleaning and processing