Designed and developed an intelligent web application that extracts YouTube transcripts and uses Retrieval-Augmented Generation (RAG) to allow users to interactively chat with video content. Engineered a multi-agent orchestration backend using FastAPI and LangChain. Implemented persistent chat sessions, timestamped bookmarks, and architected a lightweight, containerized deployment using Docker.

Built an autonomous multi-agent research system using LangGraph to search, analyze, and generate structured research reports. Developed a PDF-based RAG pipeline to ground responses with domain-specific document context. Designed a self-correcting Critic Loop to evaluate generated drafts and integrated deep web search with dynamic query expansion.

Built a multi-modal RAG chatbot capable of analyzing both text and image-heavy documents, leveraging OCR and the OpenAI API. Engineered a hybrid search and reranking engine for highly accurate, context-aware document retrieval. Containerized the application with Docker and utilized MongoDB for session management.

A modern Flask web application that uses OpenAI's GPT to provide medical information and health guidance. Features include multiple query categories, chat history, an emergency page, rate limiting, and a REST API.

A Natural Language Processing (NLP) based Spam Detection System that classifies text messages as either Spam or Ham (Not Spam) using the TF-IDF vectorizer for feature extraction and a Multinomial Naive Bayes classifier. Demonstrates a simple and effective approach to spam classification in email and SMS filtering.

Automated detection and classification of skin cancer and various skin diseases using Machine Learning (ML) and Deep Learning (DL) models. Utilizes image classification techniques to distinguish between multiple skin conditions, including Melanoma, to assist in early diagnosis. Implemented traditional ML models (Decision Tree, KNN, Random Forest, SVC) and DL approaches using MobileNet as a feature extractor.