Build with AI 2026 is a global series of community-led events designed to help developers learn and apply Google’s latest artificial intelligence technical potential.Level: Intermediate to AdvancedPrerequisites:Completion of "Building Multi-Agent Systems with ADK" workshop (or equivalent experience)Understanding of CI/CD conceptsGoogle Cloud Platform account with billing enabledFamiliarity with Python and basic DevOps practicesDeployed multi-agent system (Researcher, Judge, Content Builder, Orchestrator)Learning ObjectivesBy the end of this workshop, participants will be able to:Understand Agent Evaluation PrinciplesDifferentiate between "vibe checks" and systematic evaluationDistinguish between offline experimentation and continuous evaluationApply LLM-as-a-Judge concepts to agent quality assessmentImplement Evaluation MetricsConfigure Vertex AI evaluation metrics (adaptive rubrics, tool-use quality, hallucination detection)Create custom trajectory-based metrics for tool usageCombine code-based and model-based gradersDeploy Safe Testing InfrastructureUse Cloud Run revision tags for shadow deploymentTest agents in production without affecting usersImplement environment parity for system-level evaluationBuild CI/CD Evaluation PipelinesAutomate evaluation in Cloud BuildImplement deploy-assess-promote workflowsSet up rollback strategies for failed evaluationsSee you there!!