Constrain the output
Define the shape the application can accept, then handle the cases the model returns outside it.
CITYJS ATHENS 2026
How to Become an AI Engineer
Building Reliable AI Features in TypeScript
Build reliable AI features in TypeScript by treating the model as an engineering dependency: schema-validated outputs, evaluations in CI, evidence-based model selection, and production tracing.
CityJS CityAI Athens · 21–23 October 2026
From model call to production feature.
The workshop follows one TypeScript feature through four controls that make its behavior inspectable and repeatable.
Evaluate behavior
Put representative cases into CI so a prompt, model, or implementation change has evidence behind it.
Select with evidence
Compare models against the work the feature actually needs instead of choosing by reputation alone.
Trace production
Record enough context to understand failures, investigate changes, and improve the feature after release.
Who it is for
TypeScript engineers who can build production applications and want to move AI features beyond demos and one-off prompt experiments.
TypeScript in production
An AI feature to build or improve
A team prepared to test model behavior
Take the method back to the codebase.
The goal is not a finished demo. It is a repeatable way to constrain, evaluate, select, and observe the models inside a product.