MemoryVerse AI — Evidence-Backed Career Passport

MemoryVerse AI is an evidence-backed digital career passport that transforms scattered certificates, resumes, project reports, internship letters, achievements, and GitHub repositories into a structured and searchable professional identity. I built this project as a solo full-stack developer for MemoryVerse AI ’26. The system uses AI-powered document ingestion, OCR, metadata extraction, semantic search, embeddings, Graph-RAG, and evidence mapping to understand a student’s academic and professional journey. Unlike a conventional cloud-storage platform or portfolio, MemoryVerse AI does not merely list skills. It connects every skill to the original evidence that proves where it was learned, demonstrated, applied, or verified. Key capabilities include: • Automatic document categorization and metadata extraction • Page-aware document chunking and cited AI answers • Semantic retrieval with visible similarity and re-ranking scores • Knowledge-graph relationships between skills, projects, certifications, internships, and achievements • Confidence-based review gate that prevents uncertain documents from entering the knowledge graph • Evidence levels such as Claimed, Certified, Demonstrated, Applied, Verified, and Repeated • GitHub repository ingestion as project evidence • Visual career timeline and role-gap analysis • Secure, revocable public career-passport sharing • Visible degraded-mode warning when semantic search is unavailable Technology stack: Frontend: React, Vite, JavaScript and CSS Backend: Python, FastAPI and Uvicorn AI: Google Gemini, embeddings, OCR, NLP, semantic search, RAG and AI re-ranking Database and authentication: Supabase PostgreSQL, Supabase Auth and Storage Vector search: ChromaDB Document processing: PyMuPDF, OCR and DOCX parsing Measured development results: • 13/13 backend automated tests passed • 82% field recovery on 20 reproducibly degraded synthetic OCR scans • Correct Recall@5 retrieval for 3/3 labelled offline benchmark queries The OCR result is based on synthetic degraded test documents and is presented transparently rather than as production accuracy. GitHub: https://github.com/Jeganraj2006/MemoryVerse-AI The final outcome is a private and intelligent career repository that helps students instantly answer questions such as: “What evidence proves that I am prepared for a data analyst role?” Instead of returning unsupported claims, MemoryVerse AI provides evidence, source documents, page references, skill relationships, confidence information, and career gaps.