Atharv Taral

Production-ready LangGraph AI agent featuring Groq LLM, live search, weather, Wikipedia, web scraping, and Streamlit UI.

Production-grade Multimodal RAG Medical Assistant using LangChain LCEL for Hybrid Model Routing. Extracts handwritten prescriptions via GPT-4o-mini Vision & Pillow buffers, queries a Pinecone Serverless Vector DB (HuggingFace embeddings) to prevent hallucinations, and retains state via Streamlit.

Medical AI Assistant built with OpenAI GPT-4o-mini & Gradio 6.0. Features contextual chat memory, a professional doctor-role system prompt, and a responsive web UI. Delivers structured health advice and prescriptions using secure, modular Python logic. Ideal for GenAI portfolios.

Advanced AI chatbot system with automatic memory storage, retrieval-augmented generation (RAG), and optimized context management for long conversations.

OnionGuard is a high-precision Machine Learning solution designed to provide real-time diagnostics for onion leaf diseases. Developed as a final-year project, this system utilizes Deep Learning—specifically Convolutional Neural Networks (CNN)—to classify 11 distinct states of plant health, from fungal infections to pest damage.

Machine Learning based Insurance Premium Prediction Web App built using Lasso Regression and Streamlit.

Simple langgraph chatbot deploy in Gradio

An autonomous AI Agent built with Microsoft AutoGen & FastAPI that fetches real-time data using an async OpenWeatherMap tool. Features a clean Streamlit UI. Highly secured: all credentials, including the OpenAI API Key, are decoupled via .env and guarded by a .gitignore file.

An integrated IoT + CNN + Agentic AI system developed for precision agriculture and automated field monitoring. This project serves as a final-year major project for B.E. Computer Engineering at SPPU.

AI-powered research tool that uses multiple agents to collect data and generate reports.

A Multimodal AI-powered Medical Image Analyzer built with Streamlit and OpenAI GPT-4o. This tool interprets X-rays, MRIs, and CT scans to generate structured medical reports and features an interactive MediBot for patient-friendly discussions about the findings.

End-to-End Machine Learning project for predicting medical insurance costs using Polynomial Regression. Covers data preprocessing, feature engineering, model training, evaluation, and Streamlit deployment with an interactive UI for real-time premium prediction.