GroundedGPT — Production-Grade RAG System
Built a production-oriented Retrieval-Augmented Generation (RAG) system that enables LLMs to answer questions using retrieved information from user-provided documents. Implemented document chunking, semantic embeddings, vector search with ChromaDB, hybrid retrieval using BM25, optional reranking, and context-grounded LLM generation with citations. Developed the backend using FastAPI and integrated Ragas-based evaluation to measure answer faithfulness and relevance.
View details
Personalized Diabetes Management Platform
Built an AI-powered diabetes prediction system using TensorFlow, XGBoost, and Scikit-learn. Performed data preprocessing, feature engineering, model training, evaluation, and hyperparameter tuning. Implemented SHAP Explainable AI for transparency. Designed FastAPI REST APIs for prediction, authentication, and food image analysis. Integrated PostgreSQL for secure patient data management. Developed a Flutter frontend for real-time health monitoring.
View details
AI Mock Interview Web Application
Developed an AI-powered mock interview platform using Gemini API, Large Language Models (LLMs), and Prompt Engineering. Implemented Natural Language Processing (NLP) for intelligent response evaluation and personalized AI feedback. Built FastAPI REST APIs for interview generation, response analysis, and user management. Integrated PostgreSQL for secure interview history and analytics. Developed a responsive interface supporting multiple interview domains and AI-assisted learning.
View details
AI Wireless Network Monitor
Developed an AI-powered wireless network monitoring application for real-time packet analysis and protocol monitoring. Built an ML-based anomaly detection model using Scikit-learn for classification of abnormal network traffic. Applied data preprocessing, feature extraction, model training, and evaluation to improve anomaly detection accuracy. Designed FastAPI REST APIs with SQLite for efficient packet storage and retrieval. Built interactive dashboards using Recharts for real-time visualization of network activity.
View details
