Session 05: Teaching AI to Search
Advanced Retrieval: From Finding Information to Finding the Right Information
About the Event
Finding information is only the beginning. For an AI system to provide useful and trustworthy answers, it needs to find the right information, separate useful context from noise, and bring related pieces of information together.
In Session 05 of this series, we complete Module 2: Context Engineering & Retrieval Systems by taking our search system to the next level.
Building on the retrieval foundation from Session 04, we will explore how modern AI applications improve search accuracy, reduce irrelevant results, and handle questions that require information from multiple sources.
The session introduces practical techniques for better ranking, smarter retrieval, and connecting related information—helping us move from simply searching data to building AI systems that can retrieve meaningful context with greater precision.
Designed for campus tech students, young professionals in tech, software developers, and AI enthusiasts, this session provides a hands-on introduction to the engineering techniques behind more accurate AI search and retrieval systems. We expect approximately 40 participants, creating an interactive environment for learning, experimentation, and collaboration.
What We'll Cover
Better Search Results: Understanding why a simple similarity search isn't always enough and how retrieval systems can improve the quality of results.
Finding the Best Information: Exploring how systems can retrieve a larger set of possible answers and then identify the most relevant ones.
Connecting the Dots: Understanding how AI can combine information from different sources to answer more complex questions.
Reducing Noise: Learning how to remove unnecessary or repetitive information before it reaches the AI model.
Secure Retrieval: Ensuring that improved search does not compromise access controls or expose information users should not see.
The Hands-On Codelab: Building Advanced Retrieval
Participants will extend the search system built in Session 04 into a more accurate retrieval pipeline.
Together, we will:
Set up a vector database and load our prepared AI-ready data.
Search and retrieve relevant information using both meaning and metadata.
Build a second layer that evaluates and ranks search results by relevance.
Connect related information to answer more complex questions.
Remove unnecessary context before sending information to the AI model.
By the end of the session, participants will have a more intelligent retrieval pipeline capable of finding better information, ranking it more accurately, and providing cleaner context to AI systems.
Good AI answers don't come from searching more. They come from finding better information.