May 6

PREREQUISITES: PLEASE BRING YOUR OWN LAPTOP. This intensive 6-hour workshop provides a comprehensive walkthrough of the Databricks Data Intelligence Platform, guiding participants from foundational infrastructure to advanced Generative AI application development. Designed for data engineers, analysts, and AI developers, this session bridges the gap between traditional data engineering and modern AI agents.
Participants will begin by provisioning a secure workspace before diving into Module 2, where they will build production-grade ETL pipelines using Lakeflow Declarative Pipelines (formerly Delta Live Tables) to process complex datasets like the Pipeline Bike demo. The focus then shifts to Module 3, democratizing data access using Databricks SQL AI Functions, creating AI/BI Dashboards, and setting up Genie Spaces for text-to-SQL capabilities.
The workshop culminates in Module 4, where attendees will leverage Agent Bricks to build autonomous AI agents. This includes using ai_parse_document for extracting structured insights from unstructured PDFs (Key Info Extraction) and deploying a Knowledge Assistant (RAG) to answer domain-specific questions. By the end of the session, participants will have built a complete end-to-end intelligent data solution.

Recharge during the afternoon break — the ideal moment to enjoy a coffee, take a quick pause, or connect with fellow attendees before jumping back into the next sessions.

A conference lunch break is the perfect mix of food and conversation.
You share ideas, swap business cards, and make new connections.
By the end, you’re recharged and ready for the next session, fueled in both mind and body.

A moment to relax, recharge, and enjoy networking with others

PREREQUISITES: PLEASE BRING YOUR OWN LAPTOP WITH PYTHON DEV ENVIROMENT AND AN API KEY FROM ANY OF MAJOR LLM PROVIDERS. Development teams are experiencing a rapid increase in code contributions, many of them generated by AI tools and pair-programming copilots. As AI-generated code increases, human reviewers can quickly become overwhelmed.
The solution is to offload repetitive tasks to faster, more intelligent systems.
Sandboxes provide an ideal execution environment for this shift: isolated, reproducible, and capable of running agentic workflows safely and reliably.
By combining large language models (LLMs) and agents with existing development workflows, organizations can deploy automated reviewers that enforce coding standards, automatically fix style issues, and provide clear, actionable feedback — freeing human reviewers to focus on high-level architecture and logic.
In this workshop, Vikram Vaswani explains what sandboxes are, demonstrates how to integrate them with code review agents, and outlines strategies and best practices for deriving value from this integration.

Key Takeaways
Understand how sandboxed code review can benefit software projects
Learn how isolated sandboxes improve safety and reliability when reviewing untrusted code
See how AI agents can reduce bottlenecks and increase engineering velocity

Audience
Platform engineers and DevOps teams responsible for code quality
Engineering managers and tech leads seeking scalable review processes without overburdening human reviewers
Compliance teams requiring automated validation and enforcement of company policies and standards

Workshop Outline (180 minutes)
Introduction (5 minutes)
Bottlenecks from AI-generated code (5 minutes)
Using AI for code review (5 minutes)
Introduction to sandboxes and Blaxel (sandbox provider) — CLI, API, SDKs (15 minutes)
Hands-on live coding (140 minutes)
Create a Blaxel sandbox in the cloud
Connect to and explore the sandbox.
Execute code within the sandbox.
Create a code review agent with access to the sandbox
Connect the agent to GitHub Actions for code review on new pull requests (PRs).
Agent logging and observability
Summary and Q&A (10 minutes)