Architecture Guardrails and Observability in Agentic Software Engineering

How do you enforce and observe the quality of AI-generated code? After months of fine-tuning their custom agentic workflows, my top training groups report 100% of their code is written by AI. This talk summarizes design metrics they use as automatic feedback loops for their agents, as well as the techniques with high signal/noise they found to optimize the human code review step - the last bottleneck in software engineering.

Speaker: Victor Rentea
With 20 years of experience, Victor is an AI-Augmented Java Champion who has trained thousands of engineers from over 150 companies in a large variety of domains. The lessons learned from these sessions, he shared in countless educational talks, top-rated at major conferences, where he’s known as ‘the speaker with the soundboard’. In his spare time, Victor runs 1:1 individual coaching sessions and organizes online meetups for the European Software Crafters community. More on https://victorrentea.ro/