Field notes

Technical Writings from our Team.

Implementation notes, technical arguments, and lessons from building the Java AI stack and accountable coding systems.

Using Jev-Style Models from ModelJars

In the vast majority of RAG and Agentic applications, a lot of what the application asks of a model is to make decisions: which queue does this ticket go to, is it urgent, is this answer grounded in the…

Read article ↗

ModelJars… your local JVM AI model catalog

ModelJars publishes AI models as Maven artifacts. You add a coordinate, and the model’s weights arrive through the same resolver that brings you any other JAR resources declared in your project’s dependencies.…

Read article ↗

Deep Dive into the Vectors Library

Is the Vectors library a Vector Database? It provides persistent, searchable vector collections inside your JVM. The TLDR of it is that you can have vector collections in your JVM. Why would you want to do…

Read article ↗

A VCR for your Java AI Tests

Don’t get me wrong, unit tests are important, but integration tests give me much greater peace of mind. Add an LLM and those tests become slow, costly, tied to credentials, and unpredictable. Mocking the model…

Read article ↗

In-JVM RAG That Survives Restarts

Craig Walls published a recipe using the Vectors library that takes a working Spring AI RAG application, substitutes Qdrant for the Vectors Java Library and does away with the complexity of managing an external…

Read article ↗

The importance of code quality gates in Agentic Code Generation

Last year I was working on the idea of an agentic code generator like Claude Code that would be just as critical as I am about some engineering practices. Practices like TDD (Test-Driven Development) or…

Read article ↗

Notes tied to work we can show.

The writing program follows the lab and open-source roadmap rather than a publishing calendar. Expect implementation details, measured claims, and concrete examples.

01Java AIInference, retrieval, persistence, packaging.
02Agentic codingContext, tests, gates, evaluation.
03SystemsStorage, indexes, data structures.
04ProductsWhat changes when research ships.

Stay current

Follow the work as it develops.

Follow on Medium ↗