MemoryVerse AI '26

MemoryVerse is an AI-powered digital identity system that ingests certificates, resumes, project reports, and internship letters, then automatically categorizes, connects, and makes them instantly retrievable through natural language search and a conversational AI (Digital Twin). Setup steps: 1. Backend: cd backend, create a venv, pip install -r requirements.txt, add GEMINI_API_KEY to .env, run `uvicorn main:app --reload` 2. Frontend: cd frontend, npm install, set NEXT_PUBLIC_API_BASE_URL in .env.local, run `npm run dev` 3. Upload a few sample documents (certificate, resume, project report) to seed the knowledge graph before testing search or the Digital Twin Test credentials: single-user demo mode, no login required — all uploaded data is scoped to the local session. Key design decisions: - Every AI-derived fact (category, relationship, search result, resume bullet) traces back to its original source document — no ungrounded claims. - Search and the Digital Twin use retrieval-augmented generation with a relevance threshold, so results are scoped to what's actually relevant rather than returning everything. - Relationships are typed, not generic similarity links — Certification → Skill, Skill → Project, Project → Internship, Internship → Career Path — so the knowledge graph reflects real career progression. - The UI's growth-tree visualization unifies the knowledge graph and timeline into a single view rather than two separate screens.