Kitaru Workshop by ZenML

🌟 Want to take your agent from your laptop and transform it into a service that your platform team can effectively manage?

Many teams are either running agents locally in notebooks or are confined to a closed platform that limits them to a single model, harness, or cloud service. It's becoming clear that every organization requires its own internal agent runtime—one that operates on their preferred harness (PydanticAI, Claude Agent SDK, OpenAI Agents SDK, raw Python), can recover from crashes, has the ability to wait for human approval without burning compute resources, and executes risky tools in a secure environment.

In this engaging workshop, we will elevate a Python agent into a reliable service on real infrastructure—ensuring state persistence through crashes, enabling replay from any checkpoint, integrating human/agent approval gates, and isolated tool execution. This is designed so that platform teams can function smoothly without the need for constant oversight. We'll utilize Kitaru, the open-source runtime developed by ZenML.

💻 Bring a laptop with Python 3.10+ and get ready for an enlightening experience!

📍 Location: Center for Software Engineering Excellence (CSEE)

🎟️ Get your ticket here!