shivas

MemoryVerse AI

What I built: I developed MemoryVerse AI, an intelligent agentic platform that acts as a "digital second brain". It ingests unstructured personal data (like resumes and documents) and uses local LLMs (Ollama + LLaMA 3) to extract entities, relationships, and skills. This data is synthesized into a visual Knowledge Graph, an interactive Timeline, and actionable Career Insights. Users can also chat directly with their digital identity using RAG (Retrieval-Augmented Generation) to ask complex questions about their own experience. My Role: Solo Full-stack Developer. I architected the entire solution from the ground up, handling UI/UX design, backend orchestration, data engineering, and local LLM integration. The Stack: Frontend: Next.js (React), TypeScript, Tailwind CSS (Glassmorphism UI), Zustand Backend: Python, FastAPI, Docker, PostgreSQL, SQLAlchemy AI & Data: Ollama (LLaMA 3 + Nomic Embeddings), ChromaDB (Vector Search), spaCy (NLP), NetworkX (Graph generation) The Outcome: A production-ready, fully containerized application that processes documents in seconds, visualizes complex data relationships, and provides a completely private, offline AI Chat experience with zero API costs.