AWS Workshop: Federated Queries and Hybrid Search with Spice.ai

Amazon published a new workshop on AWS Workshop Studio that walks through deploying Spice as a data and AI substrate across AWS infrastructure. The workshop is designed for solutions architects, data engineers, and AI/ML engineers building applications and agents that need to query, search, and run LLM functions across multiple data sources.

What the workshop covers

AI applications and agents rely on a consistent set of primitives: federated query across sources, low-latency data access, hybrid search, and LLM inference. Most enterprise data systems were not built to provide all of these together, which results in development teams standing up pipelines across systems or compromising on one requirement to meet another.

This reference architecture illustrates how Spice and AWS close that gap. The workshop connects Spice to Aurora PostgreSQL, S3 Tables, S3 Vectors, and Amazon Bedrock through a single spicepod configuration deployed on EC2. From there, it guides you through:

• Federated SQL queries across sources using standard SQL
• Data acceleration through materialization and caching
• Hybrid vector and full-text search combining vector and full-text
• LLM inference invoked directly within the query engine using Bedrock foundation models

The exercises are modular, so you can work through the full workshop in two to three hours or focus on the sections most relevant to your use case.

The full workshop is available on AWS Workshop Studio.

For more details on how to build with Spice and AWS, check out the Spice.ai AWS integration page. If you have any questions or feedback, join the Spice Slack community and let us know!