DataLens โ Intelligent Data Dictionary Agent
Link to open source: https://github.com/mayankjndl/datalens
DataLens is a full-stack AI-powered data dictionary agent that connects to any database (SQLite, Snowflake, PostgreSQL) and automatically generates living, enriched documentation โ with zero manual effort.
๐ What it does:
- Auto-extracts complete schema: tables, columns, foreign keys, indexes
- Runs quality analysis grading every table A+ to F across completeness, uniqueness, and freshness
- AI enrichment via Groq (Llama 3.3 70B) writes business-ready descriptions, domain tags, and usage recommendations for every table and column
- Interactive D3 force-directed lineage graph visualises all FK relationships
- Natural language AI Chat โ ask questions in plain English, get SQL + answers instantly
- One-click Query Runner executes AI-generated SQL against real data
- โK semantic search finds any table or column by business meaning
- Exports a complete data dictionary as Markdown or JSON
๐ Built with: FastAPI ยท React ยท D3.js ยท Groq API ยท SQLite/Snowflake/PostgreSQL
๐ Demo dataset: Olist Brazilian E-Commerce (9 tables, 569,774 rows) โ Quality Score 97/100, Grade A+, fully enriched in under 2 minutes on free-tier AI
This build was uploaded as a hackathon project
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
