An AI-powered ecommerce product recommendation system with full data-scraping pipeline, PostgreSQL storage, Qdrant semantic search, and Groq Llama-3.1-8B reasoning. Supports abstract user queries, multi-step intent interpretation, and product re-ranking through a custom RAG architecture. Includes a React + Tailwind storefront, product detail pages, and an interactive chat interface with real-time AI-recommended product cards.

A time-series forecasting system for predicting mobile network traffic patterns using historical usage data. Implemented data preprocessing, feature extraction, and machine learning models to capture temporal trends and forecast future demand. Demonstrates practical applications in telecom optimization, load balancing, and smart infrastructure planning.

A bilingual (Hindi + English) AI assistant for rural governance enabling scheme lookup, Panchayat-wise data retrieval, and intelligent recommendations. Features voice input, chat history, multi-message memory, dynamic UI, and high-accuracy semantic search over government datasets using Groq LLaMA + Qdrant.

An automated SEO interlinking system with full-domain crawling, readability-based extraction, TF-IDF + SBERT hybrid similarity scoring, and intelligent anchor text generation. Includes a Streamlit dashboard with advanced controls for real-time, explainable anchor→target recommendations and CSV export.

A multilingual internship-query chatbot built with Flask, LangChain, and Groq LLaMA that converts natural-language questions into SQL for accurate internship retrieval. Supports Hindi-English seamless interaction, structured dataset search, and low-latency recommendations.

A fully offline multimodal RAG system supporting PDF, DOCX, TXT, images, and audio. Built using Flask with Tesseract OCR and Faster-Whisper ASR, e5-small-v2 embeddings, and FAISS for fast semantic search and grounded snippet retrieval.