Engineered a cutting-edge Telemedicine Assistant using LangChain, Gemini 1.5 Flash, and OpenRouter API, enabling multi-modal input interpretation (text, audio, image) for real-time diagnosis and automated medical report generation powered by the Mistral-7B Instruct model. Integrated location-aware hospital search, live doctor availability, and instant consultations via Jitsi Meet. Designed a modular, efficient backend architecture for seamless agent task orchestration and dynamic decision-making in real-time clinical scenarios..

Built an intelligent Retrieval-Augmented Generation (RAG) chatbot that delivers context-aware answers from YouTube video transcripts in real time. Leveraged Groq API with LangChain to achieve ultra-fast responses averaging 50ms latency. Developed a complete full-stack application with a Flask backend and a sleek, responsive frontend using HTML/CSS/JavaScript for smooth, interactive user experiences.

Designed and deployed an intelligent waste segregation system using dual YOLOv8 models on AWS EC2—a polythene detector that activates a rotating cutter, followed by a bio/non-bio waste classifier fine-tuned on 5K+ images with 88% accuracy. Integrated MongoDB to log bin fill levels and implemented automated alerts via Twilio WhatsApp API and buzzer when bins reach 75% capacity. Enabled semi-automated sorting using servo-controlled chambers, achieving consistent results with 1.5–2 min processing per batch across diverse single and multi-object scenarios.

Developed an OCR-powered application that extracts text from images using Tesseract OCR and OpenCV for preprocessing. Implemented functionalities like noise removal, grayscale conversion, thresholding, and text area detection to boost accuracy on low-quality images. Built an interactive Streamlit UI for users to upload images and instantly view extracted, editable text. Ideal for document digitization, scanned image analysis, and real-time text recognition use cases..

Developed an LSTM-based model for real-time detection of hand gestures representing 'A','B' and 'C,' achieving 98% accuracy. Incorporated preprocessing and data augmentation to enhance robustness, enabling low-latency translation of gestures into text for accessible communication.

IDeveloped a YOLOv8-based model for real-time classification of biodegradable and non-biodegradable waste using a dataset of 1,900+ images. Achieved high accuracy, enabling automated waste sorting to promote efficient and eco-friendly waste management.

Built a machine learning model for spam and ham email classification, leveraging natural language processing techniques. Achieved high accuracy in detecting spam emails, enhancing email filtering efficiency and user productivity.

Developed a machine learning model to predict diamond prices based on features such as carat, cut, color, and clarity. Using a dataset of diamonds, the model achieved 78% accuracy in predictions, providing valuable insights for pricing strategies in the retail sector.

Created a dynamic Power BI dashboard to analyze customer engagement trends and identify top factors impacting engagement. Adaptive visualizations and SQL integration enabled real-time insights, contributing to a 40% increase in bookings through targeted marketing strategies.

Developed using HTML, CSS, and JavaScript, this implementation supports multiplayer gameplay, allowing friends or random users to compete against each other.

This Power BI dashboard analyzes hotel booking cancellations, providing insights on cancellation rates, customer demographics, seasonal trends, and revenue impact. It enables hotel managers to identify and address common reasons for cancellations, improving retention strategies.

Developing a machine learning model for classifying waste as either dry or wet. It contains the necessary files for training and testing the model, including training images, validation datasets, and pre-trained YOLO model weights. The project aims to contribute to waste management solutions by automating the identification and classification of waste types, enhancing recycling efforts.