01fe24bci044
Featured project
Memora
What I Built: Memora — an AI-powered digital identity and knowledge graph engine that transforms scattered personal records (certificates, resumes, internship letters) into an interactive, reasoning career graph. Beyond basic document search, Memora uses a two-stage LLM pipeline to infer causal relationships (LED_TO, BUILT_ON, APPLIED_IN) between milestones, auto-generates growth narratives, and provides an interactive AI career assistant with skill gap analysis. The Stack: FastAPI (Python), Groq API (llama-3.1-8b-instant), ChromaDB, NetworkX, SentenceTransformers, Tesseract OCR, pdfplumber, Vanilla JS, Tailwind CSS, D3.js, Docker, Railway. The Outcome: Successfully built, containerized, and deployed a live multi-tenant web application globally on Railway. Achieved sub-second LLM inference, strict per-user data isolation, and optimized the Docker execution footprint to run efficiently under 200MB RAM.