Sessions

ZIO has a reputation for being powerful — but intimidating, especially if you’re not fluent in functional programming. In this talk, I explore ZIO through the lens of a machine learning practitioner who doesn’t speak in morphisms or monads — and doesn’t need to.

Why not just stick with Python? Because deploying Python ML models is notoriously painful. Scala was built for robust systems, and ZIO leverages that power to offer a practical, testable, and type-safe approach to production ML.

We’ll walk through real-world ML workflows using ZIO:

- Expressing failure and requirements with types.

- Assembling components with `ZLayer` dependency injection.

- Parallelizing training using ZIO’s concurrency model.

- Streaming large or real-time data via `ZIO Streams`.

- Serving models efficiently with `ZIO HTTP`.

If you’ve ever struggled to scale notebooks, wrangle side effects, or make your models production-ready, ZIO might be the missing piece — and you won’t need to know what a morphism is to use it.

Direct Style programming is an exciting development for the future of Scala, but something that can seem a little abstract and remote.

The purpose of this talk is to introduce the problem Direct Style is trying to solve and then, step by step, build up to the solution that the Scala compiler team is building to. We’ll see what’s already implemented and ready for use right now, but also the things that are still needed in order for the entire vision to be realised.

By the end of the talk, we’ll have created a respectable set of effects and their handlers, and written an entire program using them to demonstrate how natural that style of programming can be.