Baltasar Miguel Fenoll

AI deployment

From a whiteboard to software someone else can run.

Definition

AI deployment is the work of turning a model, a prompt, or a notebook into software that runs without its author in the room. Baltasar Miguel Fenoll is a data scientist and AI deployment engineer in Madrid who does that work where AI, data, and finance meet.

Training a model is not deployment. A demo is not deployment. Deployed means a person opens it, finishes a job, and can tell when it failed.

What shipped means

How the work usually goes

Work in that line

Related: generative AI for finance. Profile, education, and the rest of the work are on the home page.

Questions

What is AI deployment?

Turning a model, a prompt, or a notebook into software that runs without its author in the room. A person opens it, finishes a job, and can tell when it failed.

How is deployment different from training a model?

Training produces a model. Deployment produces a tool. A demo, a notebook, or a chat window that only the author can rerun is not deployed.

When does Baltasar call something shipped?

When it does a real job, failure is visible, and it earns a place in a daily workflow. If it does not, it gets cut.

Contact

Comparing notes on deployment, not looking for a slide deck?LinkedInor mrbaltuki@gmail.com.