DocuStream is an AI-driven document management platform that transforms unstructured, chaotic document repositories into a streamlined, knowledge-rich system. It automatically summarizes lengthy documents, delivers insights to the right stakeholders, and prevents compliance risks with intelligent alerts. The system enhances cross-department collaboration, preserves institutional knowledge, and eliminates duplicated effort. With real-time routing and multi-format support, DocuStream ensures faster decision-making and effortless information access for modern organizations.

Built an AI-driven queue load balancer using TensorFlow, PyTorch, and OR-Tools to predict wait times and optimize staffing. Implemented real-time crowd detection with YOLOv8 and OpenCV, developed REST APIs with Node.js and MongoDB, and deployed via Azure Container Apps. Created a responsive analytics dashboard in Next.js with TypeScript and Recharts for real-time insights.

Built a personalized AI chatbot, interview prep module, and career counselor section using Next.js, Tailwind CSS, and Firebase. Developed a resume analyzer that identifies skill gaps, recommends tailored career roadmaps, and highlights improvement areas for users.

Built a production-oriented fraud detection system for highly imbalanced financial datasets using Random Forest, SMOTE experiments, and Isolation Forest anomaly detection. Implemented threshold tuning and PR-AUC based evaluation, then exposed real-time risk scoring through a FastAPI backend. Delivered an end-to-end deployable pipeline with interpretable predictions and a frontend interface for live transaction analysis.

Developed a fully animated About page using Framer Motion, GSAP, and Lottie; implemented dynamic SEO with the Next.js Metadata API. Built a RAG chatbot with page-redirection capability powered by Gemini Flash for intelligent user query handling.