Rootconf
Rootconf

Rootconf topical edition on Databases

Call for Proposals

All submissions

Kulwant Singh

@illusivesingh

Priyansh kimtee

@pkimtee

Aniket

@aniket205

Building Enterprise‑Grade RAG Applications with SQL Server and Microsoft Fabric

Submitted Apr 29, 2026

Session type - select the format for your session: Hands-on workshops - where participants follow instructors on their laptops

Abstract:

Modern RAG applications are easy to prototype but notoriously hard to productionize. With AI agents making code generation cheap, the real challenge has shifted to understanding what breaks under real workloads—filtered retrieval, selective predicates, joins, security, live data, and unpredictable latency. This workshop aims at equipping developers with the architectural intuition needed to design robust and scalable AI systems.. Using SQL Server, we demonstrate how vector search integrated into a mature relational engine enables efficient, streaming, filtered retrieval. Participants will learn why enterprise‑grade RAG requires more than a vector store, and how a database’s optimizer, execution model, and governance capabilities provide a robust foundation for building scalable, secure, and predictable AI applications.

Key takeaways:

  • Enterprise-grade RAG with the database optimizer/execution model to make retrieval efficient under filters, joins, and streaming constraints.
  • Treat vector search as an access path inside a governed data platform, with entrprise security, reliability, and live updates etc.

Who will benefit from this session?

  • Application developers and platform engineers building production RAG/agentic apps who need predictable latency, filtered retrieval, and secure grounding.
  • Data engineers and architects using Microsoft Fabric/OneLake who want an end-to-end pattern for ingestion → chunking/embeddings → serving with SQL.

Presenter info:

The presenters are experienced software engineers at Microsoft based in Bengaluru, working in the SQL engineering organization. Their work focuses on SQL platform capabilities and performance-oriented patterns for intelligent applications, including vector search and scalable retrieval architectures.

Outline

Module 0: Setup & Environment Check (10 min)

Module 1: Embeddings & Vector Storage (15 min)

Generate vector embeddings and create a vector index.

Module 2: Vector Search (15 min)

Vector search in Azure SQL.

Module 3: RAG API — GraphQL + Chat Completion (20 min)

Create and deploy API in Fabric.

Module 4: Production Patterns (15 min)

Integerate to have real world scenarios with joins, filters, securitcy etc.

Module 5: Wrap-up & Next Steps (10 min)

Q&A.

Prerequisites

Below would be nice to have

  • Basic SQL (SELECT, JOIN, stored procedures, DDL)
  • Conceptual understanding of REST APIs
  • Familiarity with cloud services (Azure portal basics)
  • Understanding of embeddings & vector similarity
  • Experience with Azure OpenAI or any LLM API
  • Familiarity with Microsoft Fabric

Instructions page

Link to Instructions @ GitHub

All submissions

Comments

{{ gettext('Login to leave a comment') }}

{{ gettext('Post a comment…') }}
{{ gettext('New comment') }}
{{ formTitle }}

{{ errorMsg }}

{{ gettext('No comments posted yet') }}

Hosted by

Rootconf
Rootconf
We care about site reliability, cloud costs, security and data privacy