Operations | Monitoring | ITSM | DevOps | Cloud

In this episode, Julien Simon, longtime voice in the open-source ML world, reminds us that even in the era of GenAI, reliability fundamentals haven’t changed. Julien breaks down why calling “the same model” from different providers can produce wildly different results, how deployment choices introduce hidden variability, and why reliability teams need to think of LLM systems as distributed systems.

Rootly is the AI-native on-call and incident management platform that helps you resolve incidents faster, improve system resilience, and streamline on-call operations. It’s your always-on SRE copilot that automates root cause analysis and identifies patterns that drive continuous improvement—trusted by thousands of companies like LinkedIn, NVIDIA, Replit, Elastic, Canva, Clay, Tripadvisor, and Grammarly.

Only 50% of companies monitor their ML systems. Building observability for AI is not simple: it goes beyond 200 OK pings. In this episode, Sylvain Kalache sits down with Conor Brondsdon (Galileo) to unpack why observability, monitoring, and human feedback are the missing links to make large language model (LLM) reliable in production.