add datathon documentation

Author: paladique

.devcontainer/devcontainer.json

@@ -0,0 +1,8 @@
+{
+  "name": "SQL + AI Contest",
+  "image": "mcr.microsoft.com/devcontainers/python:3.11",
+  "features": {
+    "ghcr.io/devcontainers/features/docker-in-docker:2": {}
+  },
+  "forwardPorts": [8000, 5000]
+}

.github/workflows/mission1-embeddings-search.yml

@@ -0,0 +1 @@
+# Mission 1 workflow placeholder

.github/workflows/mission2-dab-app.yml

@@ -0,0 +1 @@
+# Mission 2 workflow placeholder

.github/workflows/mission3-agent-rag.yml

@@ -0,0 +1 @@
+# Mission 3 workflow placeholder

.github/workflows/submission-final.yml

@@ -0,0 +1 @@
+# Final submission workflow placeholder

README.md

@@ -1 +1,38 @@
-# sql-ai-contest-starter
+# SQL AI Datathon Starter Repository
+
+Welcome to the SQL AI Datathon! You'll complete guided missions that will prepare you for building AI-powered applications with Microsoft SQL. You’ll use AI features to build embeddings, semantic/hybrid search, DAB endpoints, and an agent with RAG.
+
+Complete all four missions to earn a badge and unlock the Open Hack challenge.
+You'll add your own creative twist in the Open Hack and be eligible for judging for a ticket to SQLCon 2026!
+
+
+## Quick Start
+1. Fork this repo.
+2. Open in Codespaces or local devcontainer.
+3. Run `scripts/...` to populate sample data.
+4. Commit a change to trigger **Mission 1** CI; fix any failing tests.
+5. Repeat for Missions 2 & 3. Use Week 4 to improve the app.
+
+## Mission Briefs
+1. [Embeddings & Search](missions/mission1/README.md)
+2. [Retrieval Augmented Generation (RAG)](missions/mission2/README.md)
+3. [Orchestrate SQL + AI workflows in C# or Python](missions/mission3/README.md)
+4. [Building a Full-Stack AI Application in C# or Python](missions/mission4/README.md)
+7. [Open Hack]()
+
+## What you’ll build
+- Embeddings & search 
+- REST/GraphQL endpoints with Data API Builder
+- An agent that answers questions grounded in your SQL data
+
+## How you’re evaluated for the Open Hack
+- CI must pass for each mission to receive a badge and your Open Hack submission to be eligible for judging
+- README clarity & reproducibility
+- Creativity, impact, and alignment with SQL + AI objectives
+
+# How to Submit
+1. Ensure all mission workflows passed on your repo.
+2. Open an Issue on [this repo](https://github.com/your-repo-link) using the “Final Submission” template.
+3. Paste links to:
+   - Latest successful workflow runs
+   - Repo README and any demo instructions

missions/mission1/README.md

@@ -0,0 +1,17 @@
+# Mission 1: Creating Embeddings and Performing Search
+
+You will be guided through implementing semantic search capabilities using embedding models and Azure SQL Database. In this mission, you will:
+
+## Learning Objectives
+- **Convert Text to Vectors**: Use an embedding model to convert text into high-dimensional vector representations
+- **Store Embeddings**: Store embeddings efficiently in Azure SQL Database
+- **Query with Vector Similarity**: Query the database using vector similarity to find semantically related content
+- **Maintain Embeddings**: Keep embeddings updated as data changes
+
+## Prerequisites
+1. Azure SQL Database with populated data
+1. Embedding model access (you can use the free AI Proxy to generate embeddings for free)
+
+## Next Steps
+After completing this mission, you will have implemented a semantic search solution that can find relevant information based on meaning rather than just keywords.
+Proceed to [Mission 2: Retrieval Augmented Generation (RAG)](../mission2/README.md) to learn how to build an end-to-end RAG pipeline that combines embeddings with language model generation.

missions/mission2/README.md

@@ -0,0 +1,25 @@
+# Mission 2: Retrieval Augmented Generation (RAG)
+
+You will be guided through implementing retrieval augmented generation (RAG) capabilities using embedding models and Azure SQL Database. In this mission, you will:
+
+## Learning Objectives
+- **Implement RAG Pipeline**: Build an end-to-end retrieval-augmented generation workflow
+- **Query with Context**: Use vector similarity search to find relevant documents based on user questions
+- **Generate Informed Responses**: Feed retrieved context to a language model to produce accurate, grounded answers
+- **Prevent Hallucinations**: Learn techniques to ensure LLM responses are based on actual data rather than fabricated information
+- **Optimize Context Windows**: Understand how to select and rank the most relevant information for the language model
+
+
+## Prerequisites
+1. Mission 1 completed
+1. Embedding model access (you can use the free AI Proxy to generate embeddings for free)
+1. Language model access (you can use the free AI Proxy to generate completions for free)
+
+# Key Concepts
+This mission demonstrates how to build production-ready AI applications that combine the semantic understanding of embeddings with the generative capabilities of large language models. RAG is a fundamental pattern for creating trustworthy AI systems that answer questions based on your specific domain knowledge.
+
+The techniques learned here enable you to build chatbots, knowledge assistants, and intelligent search systems that provide accurate, source-backed responses.
+
+## Next Steps
+After completing this mission, you will have implemented a robust RAG pipeline that can answer questions based on your SQL data.
+Proceed to [Mission 3: Orchestrate SQL + AI workflows](../mission3/README.md) to learn how to build complex, multi-step AI workflows that integrate RAG with other SQL and AI capabilities.

missions/mission3/README.md

@@ -0,0 +1,24 @@
+# Mission 3: Orchestrate SQL + AI workflows
+
+This mission advances your skills by teaching you how to orchestrate complex workflows that combine SQL database operations with AI capabilities. You'll learn to build intelligent, multi-step processes that seamlessly integrate data retrieval, transformation, and AI-powered decision making.
+
+## Overview
+Learn how to design and implement sophisticated AI workflows that coordinate multiple database queries, embedding operations, and language model calls to solve complex business problems requiring multiple steps and decision points.
+
+## Learning Objectives
+- **Build Multi-Step Workflows**: Create orchestrated processes that chain together SQL queries, vector searches, and AI model calls
+- **Implement Conditional Logic**: Use AI responses to determine next steps in your workflow
+- **Handle Data Transformations**: Process and prepare data between workflow stages
+- **Manage State and Context**: Maintain information across multiple operations in a workflow
+
+## Prerequisites
+1. Mission 1 completed (embeddings and vector search)
+1. Mission 2 completed (RAG implementation)
+1. Understanding of async/await patterns
+1. Embedding and language model access
+
+
+## Next Steps
+After completing this mission, you will have built a robust AI orchestration layer that can handle complex scenarios by integrating SQL data operations with AI capabilities.
+
+Proceed to [Mission 4: Building a Full-Stack AI Application](../mission4/README.md) to learn how to create a complete application that leverages the workflows you've built in this mission.

missions/mission3/dotnet/mission3.ipynb

@@ -0,0 +1,43 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ce33247c",
+   "metadata": {
+    "vscode": {
+     "languageId": "plaintext"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "# Mission 3: Orchestrate SQL + AI workflows\n",
+    "\n",
+    "This mission advances your skills by teaching you how to orchestrate complex workflows that combine SQL database operations with AI capabilities. You'll learn to build intelligent, multi-step processes that seamlessly integrate data retrieval, transformation, and AI-powered decision making.\n",
+    "\n",
+    "## Overview\n",
+    "Learn how to design and implement sophisticated AI workflows that coordinate multiple database queries, embedding operations, and language model calls to solve complex business problems requiring multiple steps and decision points.\n",
+    "\n",
+    "## Learning Objectives\n",
+    "- **Build Multi-Step Workflows**: Create orchestrated processes that chain together SQL queries, vector searches, and AI model calls\n",
+    "- **Implement Conditional Logic**: Use AI responses to determine next steps in your workflow\n",
+    "- **Handle Data Transformations**: Process and prepare data between wor\n",
+    "- **Manage State and Context**: Maintain information across multiple operations in a workflow\n",
+    "kflow stages\n",
+    "\n",
+    "## Prerequisites\n",
+    "1. Mission 1 completed (embeddings and vector search)\n",
+    "1. Mission 2 completed (RAG implementation)\n",
+    "1. Understanding of async/await patterns\n",
+    "1. Embedding and language model access"
+   ]
+  }
+ ],
+ "metadata": {
+  "language_info": {
+   "name": "python"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}