Devoxx: Beyond the Prompt: Evaluating, Testing, and Securing LLM Applications by Mete Atamel

Beyond the Prompt is Mete Atamel’s deep dive into how we really test and lock down our LLM apps. Instead of just riffing on prompts, he shows how to track tweaks in prompts or RAG pipelines and plug in frameworks like Vertex AI Evaluation, DeepEval, and Promptfoo to keep your models honest.

On the security side, he introduces LLM Guard to fend off sneaky prompt injections and block harmful outputs. The upshot? You need solid input–output guardrails to make sure your LLM never goes off the rails.

Watch on YouTube