2025-06-12 –, D105 (capacity 300)
So, you’ve got a new colleague: an AI agent. It doesn’t need coffee breaks, doesn’t complain about on-call schedules, and (so it claims) can debug faster than you. Sounds perfect, right? Well, until it starts spamming Slack at 3 AM or, worse, decides it has production access opinions.
In this session, we’ll dive into the chaotic, glorious reality of managing AI agents as part of your SRE team. We’ll tackle the tough questions: How do you assign permissions to something that thinks it’s always right? How do you monitor and log what your robot coworker is doing (and why it thought deleting all your service accounts was a good idea)? And most importantly, how do you keep it from going rogue and bringing down prod on a Friday night?
With a mix of traditional SRE practices and new strategies tailored for AI agents, we’ll explore how to establish trust, maintain control, and avoid turning your AI into an unmanageable production liability. By the end of this talk, you’ll have the tools to SRE your AI colleague—because we all know it’s not a real teammate until it’s on-call with you.
Intermediate - attendees should be familiar with the subject
I graduated from Computer Engineering in 2015 and began my career in Istanbul, working as a Software and DevOps Engineer across Banking, Media, and E-commerce industries.
In 2019, I joined Red Hat as a Senior Site Reliability Engineer on the Platform team, focusing on scaling and managing the complexities of Red Hat’s Managed OpenShift offering. Now part of the Hybrid Cloud Platforms Security team, I combine my experience in reliability engineering with a passion for security, tackling challenges like securing production systems, safeguarding cloud-native environments, managing policy enforcement, and, of course, figuring out how AI can play nice in cloud-native environments without breaking things (too much...)