I am a Ph.D. scholar at the Department of Computer Science & Engineering, IIT Kanpur. My work focuses on Generative AI and Retrieval-Augmented Generation (RAG), with interests in LLM fine-tuning, efficiency, and evaluation.
Multi-PDF chat app with GROQ (Llama-3.1-70B), LangChain chunking, FAISS retrieval, Instructor-XL embeddings, and conversational memory. Built in Streamlit for one-page UX.
ML-based detector that analyzes URL features (IP presence, length, redirects, HTTPS, domain age/WHOIS, traffic metrics) to flag phishing attempts. Streamlit front-end.
Emotion recognition with CNNs to classify facial expressions as happy vs. sad; trained and evaluated for accuracy and generalization.
Supervised ML model to forecast home prices based on location, size, amenities, and historical trends; includes data cleaning and feature engineering.
Enhanced RAG framework that turns documents & websites into interactive knowledge bases with ChromaDB, HuggingFace embeddings, and GROQ-powered LLM; handles up to 10 sources with multi-turn context.
Conversational interaction with any website using GROQ (Llama-3.3-70B), vector storage, and conversational memory for context-aware answers.
Email: uditanshu25@iitk.ac.in
GitHub: github.com/UditanshuPandey
LinkedIn: linkedin.com/in/uditanshupandey
