SQL + AI Datathon

This is a reminder!

A ticket to FabCon or SQLCon is on the line — and winners will be featured in a conference session.

Register and join us for the livestream series on February 12th and 19th where SQL experts will share tools and tips to help you be a proficient AI engineer with SQL and walk you through the datathon missions.

How to participate in the datathon:
Our first-ever SQL + AI Datathon is built to walk you from zero to a working AI-powered SQL solution - step by step.
The Datathon officially launches tomorrow and runs for three weeks, with submissions closing on February 26 at 11:59 PM PT. Everything you need — missions, setup guides, starter code, and the final challenge — is in the GitHub repository.

1. Start with the Missions (Your Guided On‑Ramp)
The Missions folder in the repo is your starting point. Work through Missions 1–4 to build the core components you’ll need for the final challenge.

• Mission 1: Create your database, load the sample data, and set up your chat models (either using your own setup or the free proxy when it becomes available).
• Mission 2: Learn how to use embeddings + the chat model together for Retrieval-Augmented Generation (RAG). You’ll turn user questions into embeddings, retrieve relevant rows from SQL, and produce structured outputs.
• Mission 3: Choose your path — Python or .NET — and follow the notebook to build a small AI agent that can answer user queries using your SQL-backed RAG flow.
• Mission 4: Build a simple full-stack app (Python or .NET) that can search the database, page through results via Data API Builder, and provide a lightweight chat experience.

By the end of the missions, you’ll have all the building blocks needed to confidently tackle the final challenge.

2. Complete the Final Challenge
After finishing the missions (or as many as you need), move to the Open Hack section in the repo.
Choose one project type:

• Chatbot Agent, or
• Semantic Search Tool

Your solution must:

• Use SQL as the data source,
• Implement RAG (retrieve from SQL + use AI to enhance responses),
• Include documentation showing how your solution works and how someone else can reproduce it.

3. Submit Your Entry
When you're ready, submit your project using the instructions in the repository. We’ll collect all entries directly from the GitHub Issues template—so no extra forms or portals.

4. Judging
Entries will be evaluated on:

• Problem Understanding & Relevance
• Innovation & Creativity
• Technical Execution & Reproducibility
• Insight & Impact

(Full judging criteria are listed in the repo.)
Please read the official contest rules.

Contact us at: UttarPradesh.PowerBIClub@msftcommunity.com