IDUG Brazil 2026

Production-ready AI applications don't happen by accident. They require well-architected, scalable, and maintainable data and AI pipelines. In this hands-on workshop, participants will build the two pipelines that most production AI applications depend on, using IBM Db2's modern AI stack, covering native vector storage, approximate vector indexing, and built-in integration with external AI models.

Participants will first build a Data Ingestion Pipeline that loads, processes, and transforms source content into searchable vectors, leveraging external AI models registered and called directly through Db2. They will then build a Search and Retrieval Pipeline that finds the most relevant data for any given query using Db2's native vector search capabilities, and delivers the most relevant results back to the application accurately and efficiently.

DBAs will recognize familiar ground here. The same principles that make a great database — performance, scalability, security, and maintainability — are exactly what separate a production AI application from a proof of concept. For AI engineers and application builders, this workshop offers something equally valuable: a solid, enterprise-grade data foundation that eliminates the complexity of managing separate vector stores and AI infrastructure, so they can focus on building intelligent applications faster.

Participants leave with hands-on coding experience building production-grade pipelines, along with the practical knowledge to tune for performance, scale for large workloads, and maintain AI applications with confidence over time.

Familiarity with SQL and basic database concepts is all that is needed.