Yash Pandey
Featured project
MemoryVerse AI '26 || Everbrain
Everbrain is an AI system that turns the scattered evidence of a person's academic and professional life — certificates, resumes, project reports, internship letters, repositories, articles, portfolio pages — into a single structured, searchable knowledge graph, without ever altering the original files. It is explicitly not cloud storage. Storage can hold a certificate and a project report in the same folder and still not know that the certificate is why the project was possible. Everbrain reads each document, classifies it, extracts the skills / projects / credentials inside it, resolves duplicates across sources, connects related things with an explanation, and answers questions in natural language — while every answer links straight back to the untouched original. Everything you need architecture, pipeline, API reference, setup steps, and design decisions is in the README (with the thought process in THOUGHT_PROCESS.md). The project is already hosted at https://digiident.vercel.app — no setup needed. Note: the backend runs on a free tier and cold-starts, so the first request may take 2–3 minutes to wake up; after that it's instant. No account needed to explore - click "See it working first" on the landing page for a fully ingested sample record. To test with your own data, create an account via "Build your brain". The UI is fully hand-designed no component library: a seven-colour category system used consistently across every surface, a live force-directed graph, exportable highlight cards, and a GitHub-style dark mode. A large share of the build time went here every animation, folder, and card was deliberately crafted rather than templated.