Ashish Choudhury

MemoryVerse AI '26

Reviewer Note Thank you for reviewing MemoryVerse AI. MemoryVerse AI is designed to solve a common problem faced by students: important academic and professional documents are often scattered across multiple folders, making them difficult to organize, search, and connect. This project transforms those disconnected files into an intelligent, searchable digital identity while always preserving the original evidence. Recommended Demo Flow For the best experience, please follow these steps: 1. Launch the application using `start_memoryverse.bat` (Windows) or follow the manual setup instructions. 2. Upload multiple documents together (certificates, resumes, project reports, internship letters, portfolios, etc.) through Add Evidence. 3. Wait for the indexing process to complete. 4. Open Ask Your Vault and try questions such as: Which documents prove my machine learning experience? Show all my internships. What certifications do I have? Which projects are related to Python? 5. Explore the Connections page to see how projects, internships, certifications, and achievements are linked together. 6. Open the Timeline to view the automatically generated academic and professional journey. 7. Visit the Evidence Library to verify that every AI-generated insight is backed by downloadable original documents. Key Features to Evaluate Intelligent document ingestion from multiple formats. Automatic categorization of evidence. Semantic search across all uploaded files. Explainable relationship graph between achievements. Grounded AI answers with source citations. Preservation of untouched original documents. Fully local-first architecture with no mandatory cloud services. Technical Highlights FastAPI backend with Streamlit frontend. SQLite for structured metadata. ChromaDB for semantic retrieval. NetworkX knowledge graph. Local deterministic embeddings (no model download required). Optional Ollama integration for enhanced AI extraction. Notes The application works completely offline by default. No API keys or cloud accounts are required. Ollama integration is optional and automatically falls back to the local pipeline if unavailable. Thank you for taking the time to evaluate MemoryVerse AI. We hope the project demonstrates how AI can transform scattered student documents into an organized, explainable, and searchable digital identity while maintaining privacy and preserving every original piece of evidence.