Ashwin Jauhary
Featured project
MemoryVerse AI
🔗 Source Code (GitHub): https://github.com/Ashwinjauhary/MemoryVerse-AI Students build incredible digital footprints over the years, but their certificates, resumes, and projects end up scattered across drives and emails. MemoryVerse AI solves this by acting as a spatial, AI-powered digital identity archive. You simply drag and drop your scattered files, and our AI pipeline automatically reads, categorizes, and structurally connects your entire life journey into a beautifully organized, searchable timeline. Key Features: • Zero-Touch Ingestion Engine: Upload any PDF or Image. Uses Tesseract OCR to extract text and Groq (Llama 3) to automatically generate summaries, exact dates, and categories without manual data entry. • Semantic Relationship Engine: Powered by sentence-transformers and Supabase pgvector, the AI analyzes semantic meaning and automatically links related milestones together (e.g., connecting a React course to a Frontend Developer role). • Chronological Timeline: A beautifully crafted, glassmorphic UI that weaves unstructured data into a continuous, color-coded timeline of your growth. • Smart Semantic Search: Search for concepts (e.g., "web development experience") and the AI will retrieve relevant projects and certificates based on vector similarity, even without exact keyword matches. Tech Stack: • Frontend: Next.js (React), Tailwind CSS, deployed on Vercel. • Backend: FastAPI (Python), Tesseract OCR, PyTorch, deployed on Render. • AI/ML: Groq (Llama-3.3-70b-versatile) for NLP, HuggingFace (all-MiniLM-L6-v2) for embeddings. • Database & Auth: Supabase (PostgreSQL + pgvector) and GitHub OAuth.
