Jarvis AI Dictionary Agent
Link to open source: https://github.com/narang25/intelligent-data-dictionary-agent
JARVIS is an AI-powered Intelligent Data Dictionary and SQL Copilot designed to transform raw database schemas into interactive, conversational intelligence.
The system automatically documents database tables, extracts relationships, generates embeddings using Sentence Transformers, and enables Retrieval-Augmented Generation (RAG) for accurate contextual responses.
It allows users to:
• Explore database schemas in natural language
• Identify data quality issues using profiling metrics
• Generate safe, validated SQL queries
• Execute analytical queries securely
• Maintain conversational memory across sessions
• Auto-document large schemas through background ingestion pipelines
The platform is built using FastAPI, PostgreSQL (with pgvector), Redis, Celery, Docker, and Sentence Transformers. It supports vector similarity search, schema context building, and enterprise-grade SQL validation layers.
By combining AI reasoning with structured schema grounding, JARVIS acts as a data intelligence assistant for analysts, engineers, and decision-makers.
This build was uploaded as a hackathon project








