Ayushi Tripathi

specializing in content analysis through NLP and generative AI

Completed a Computer Vision project on document intelligence using AI and NLP, including developing an image processing tool with OpenCV to enhance OCR accuracy and implementing NER for efficient extraction of key information from legal contracts. Leveraged BERT, a pre-trained NLP model, to perform text preprocessing, semantic vectorization, and named entity recognition tasks. Tokenized raw text data into subwords, generated context-aware word embeddings, and fine-tuned the model on annotated data for efficient named entity recognition. Developed a machine learning model to categorize expenses from a spend dataset into various spend categories, using Random Forest, Decision Trees etc. This solves the common challenge of manual expense categorization, which is time-consuming and error-prone.

Implemented YOLO, a real-time Computer Vision object detection algorithm that utilizes convolutional neural networks (CNN), specifically for detecting weapons and vehicles. Dockerized the Streamlit app for the object detection project, allowing packaging and running the application in a secure and isolated environment. Leveraged Kafka for real-time data streaming in an image search engine, enabling high-performance processing of large volumes of data.

Received a Director medal for securing highest grade point in bachelor's.