MediQuery – AI Medical Document Assistant
AI clinical assistant that lets patients query their medical records in plain English using a full RAG pipeline — with semantic retrieval, multi-turn memory, and cited answers.
Initializing...
Creating intelligent AI products that learn, reason, and automate. Focused on Machine Learning, Agentic AI, LangChain, RAG Systems, and full-stack AI applications built for real-world impact.

LangChain + RAG
AI Engineer focused on building AI agents, chatbots, and LLM-powered applications using modern machine learning technologies.
I specialize in building end-to-end AI solutions from research to production deployment.
Build autonomous AI agents, intelligent chatbots, and workflow automation systems using LangChain, RAG architectures, and modern LLM frameworks.
Develop machine learning solutions for prediction, classification, and intelligent decision-making through data-driven models and modern ML techniques.
Build end-to-end AI products by combining modern frontend technologies with scalable backend systems, databases, APIs, and LLM integrations to deliver real-world intelligent solutions.
Technologies and tools I use to build AI agents, machine learning solutions, and full-stack AI applications.
A showcase of AI Agents, RAG Systems, and Full-Stack AI Solutions.
AI clinical assistant that lets patients query their medical records in plain English using a full RAG pipeline — with semantic retrieval, multi-turn memory, and cited answers.
An enterprise-grade AI platform that transforms customer feedback into actionable business insights using fine-tuned BERT models, FAISS vector search, semantic retrieval, intent classification, and Gemini-powered analysis.
A multi-agent research platform that autonomously researches topics, gathers information from the web, generates structured reports, and evaluates report quality through specialized AI agents.
An AI-powered chatbot built with LangGraph and LangChain featuring multi-model support, agent workflows, optional web search, and a FastAPI backend with Streamlit frontend.
A multi-modal AI coaching system that analyzes speech clarity, filler words, body language, and sentiment in real-time using computer vision and natural language processing.
An autonomous safety monitoring and voice companion system that detects falls, provides health reminders, and automatically alerts caregivers during emergencies.
A localized medical AI assistant that combines Gemini reasoning with healthcare resources, hospital discovery, medicine lookup, and symptom analysis.
Pursuing a degree in Computer Science with a focus on AI and Machine Learning.
Completed intermediate education with a focus on pre-engineering.
Completed a structured AI/ML program building and evaluating classification and regression models using Python, scikit-learn, pandas, and NumPy across end-to-end ML pipelines.
Applied industry best practices in data preprocessing, feature engineering, model evaluation, version control, and results reporting — preparing production-ready AI/ML workflows.
Developed prototypes to simulate model training, testing, and deployment scenarios, gaining practical experience with pickle serialization and real-time inference.
I'm always open to discussing new projects, creative ideas, or opportunities to be part of your vision. Feel free to reach out through any of the platforms below.