QBIT — Federated AI Analytics Platform

Overview

An enterprise-grade Federated AI Analytics Platform that lets users query disparate data sources (Snowflake, Databricks, PostgreSQL, and others) using natural language. Instead of waiting for data engineering teams to write complex SQL joins, users ask high-level business questions. A multi-agent orchestration pipeline and Spice.ai query acceleration decompose queries, map them to underlying schemas, execute federated queries in real time, and return interactive dashboards, executive summaries, and downloadable PDF/PowerPoint decks.

Tech stack

• FastAPI
• Python
• PostgreSQL
• SQLite
• OpenAI GPT
• Snowflake
• Databricks
• Spice.ai
• SSE
• Auth0
• AES-256
• PowerPoint / PDF generation

Features implemented

• Natural language querying across enterprise data sources
• Federated data layer for Snowflake and Databricks without manual ETL
• Multi-agent AI pipeline for query decomposition, semantic understanding, SQL generation, and validation
• Real-time streaming of AI responses using Server-Sent Events (SSE)
• Dynamic dashboards and executive insights for business reporting
• Automated PowerPoint and PDF report generation from AI-generated insights
• Secure connector management with AES-256 encrypted credentials
• Auth0 authentication and role-based access control
• Automated metadata discovery and schema synchronization for connected sources

Key achievements / impact

• Enabled business users to query federated enterprise data without waiting on custom SQL joins
• Built a production multi-agent orchestration pipeline with real-time streaming responses