Jeff Huber is the co-founder and CEO of Chroma, which builds infrastructure for retrieving information for AI applications. He writes about context engineering: selecting and testing the information a model receives, rather than relying on a larger context window alone.[1][2][3]
Start with The Rise of Context Engineering →
X Career
Contributions
His Chroma interview explores the path from an embedding database to infrastructure for useful AI applications. The practical concern is how developers connect their own information with a model and evaluate the result. Retrieval becomes part of the product’s behavior, rather than an interchangeable step hidden behind a chat interface.[2]
Ideas
Huber’s context-engineering essay separates gathering potentially relevant information from selecting what belongs in the final context. Search can maximize recall; reranking and other selection methods improve precision. He emphasizes testing changes against query–information pairs so developers can measure whether a different embedding model, chunking strategy or reranker actually helps their application.[3]
Tools, writing & talks
Writing, research & talks
Start here
The Rise of Context Engineeringjeffhuber.substack.comHuber’s explanation of gathering, selecting and evaluating information for an AI application.
Read the Gather and Glean distinction, then the section on a golden dataset of query–information pairs. Choose a small retrieval task and define the relevant information before comparing embeddings, chunking or reranking.
Author[3]A Latent Space discussion of retrieval infrastructure and building applications around models and private data.
Interview guest[2][7]Original source