Memoryverse Ai

InspirationStatic resumes and scattered certificates fail to show the true depth of real-world experience.

We wanted to bring career artifacts to life through an interactive, verifiable AI digital twin.

What it doesIngests documents and certificates to automatically extract skills, milestones, and domain expertise.

Uses neural RAG and Gemini to let users, recruiters, or peers converse with an AI twin that cites exact evidence.

How we built itBuilt with a React 18 & Tailwind CSS frontend and an Express TypeScript backend on Google Cloud Run.

Powered by Google Gemini (@google/genai) for multi-step entity extraction and vector embeddings for semantic search.

Challenges we ran intoEliminating hallucinations to ensure the AI twin only speaks from verified, uploaded documents.

Normalizing messy, unstructured multi-page PDFs into clean, queryable knowledge chunks.

Accomplishments that we're proud ofBuilt a high-accuracy RAG pipeline with real-time, document-level source citations.

What we learnedHow to combine structured JSON outputs with Gemini for deterministic resume and skill parsing.

Best practices for hybrid search (dense embeddings + keyword filters) to maximize retrieval precision.

What's next for Memoryverse Ai Real-time voice-to-voice mock interview simulations tailored to your uploaded vault.

Direct GitHub repository syncing and automated ATS-optimized resume exports.

Built With

• llm
• node.js
• rag
• react
• tailand
• typescript