Memex is a personal knowledge base that ingests your social media exports — LinkedIn, Twitter/X, Instagram, PDFs, images — and lets you have a conversation with them. Ask a question and get a grounded, cited answer pulled from your own words, scoped privately to your account.

Tech Stack

Frontend — React 19, Vite, Tailwind CSS v4, Lucide React
Backend — Node.js, Express 4, ES Modules
Auth — GitHub OAuth + email/password (bcryptjs, JWT session cookie via jsonwebtoken + cookie-parser)
Vector DB — Qdrant Cloud (@qdrant/js-client-rest), per-user isolated collections
Embeddings — OpenAI / Gemini (@google/genai) / local deterministic fallback (no API key needed)
LLM — Groq / OpenAI / Gemini, with automatic priority fallback
Parsers — csv-parse (LinkedIn), cheerio (Instagram HTML), pdfjs-dist (PDF), tesseract.js (image OCR, dev only)
Deployment — Vercel (serverless functions + static client

What Each Layer Does

Auth — signs in via GitHub or email/password, issues a signed session cookie; every request downstream carries a userId.
Ingestion — parses uploaded files by platform, splits them into chunks, skips ones already seen (content-hash dedup), embeds the rest, and stores them in Qdrant tagged with the owner's userId.
Retrieval — embeds the incoming question and does a vector search filtered to the current user's own data — no cross-user leakage.
Answer Service — feeds the retrieved chunks to an LLM with a "cite your sources" system prompt, returning an answer plus per-chunk citations.
Frontend — a single chat view: ask questions, attach files via the composer's paperclip icon, see cited sources in the side panel.