Prompting Safely: Building Secure LLM Prompts with a Scala DSL - Ignacio Gallego Sagastume & Isaias Bartelborth
Lambda World 25
•
36m
Discover how a domain-specific language (DSL) in Scala can help you design secure and composable prompts for LLMs. This talk explores techniques to prevent prompt injection, enforce prompt structure, and reason about non-deterministic behavior using a monadic effect system and Functional Programming. Also, we will show how to use automatic serialization libraries (like Circe, Play or ZIO Schema) and Scala case classes to model LLM structured responses in JSON format.
Up Next in Lambda World 25
-
Does FP still matter in the era of AI?
Closing debate at Lambda World.
-
Verse - A New Functional Logic Langua...
Verse is a functional logic language mainly designed by Tim Sweeney at Epic Games. It has several unique features. In this talk I will briefly present the language, and dive a little deeper into the novel way function definitions work.
-
From Functions to Inlinable Effects i...
In this talk, we'll explore how Kotlin combines imperative and functional programming styles. We'll look at how to manage effects without relying on effect handlers, how inline functions play a key role, and how must-use return values help gently distinguish pure functions from those with side ef...
