Meetup Panel: When AI Meets Production: What Breaks, What Scales, and Engineering AI That Actually Works
Meetup Panel: When AI Meets Production: What Breaks, What Scales, and Engineering AI That Actually Works
Building an impressive AI prototype is one thing. Running an AI system reliably in production—across thousands or millions of interactions—is an entirely different engineering challenge.
In this panel, experienced technology and engineering leaders will go beyond the AI demo to discuss what actually happens when AI systems meet real-world production environments. The discussion will examine the organizational side of production AI: how engineering, platform, data, security, and product teams establish ownership and guardrails without slowing innovation.
The conversation will also explore the challenges teams encounter as they move from experimentation to enterprise deployment, including hallucinations and unpredictable model behavior, latency and performance bottlenecks, observability gaps, evaluation and testing, reliability, incident response, infrastructure complexity, security, and rapidly growing inference and operating costs.
Panelists will share practical lessons from building, deploying, and maintaining production AI systems, including how their teams monitor AI applications, determine whether models are performing as expected, respond when systems fail, and make decisions about models, infrastructure, and architecture as usage scales.
Attendees will leave with a clearer understanding of what it takes to move beyond successful prototypes and build AI systems that are observable, reliable, cost-conscious, and capable of operating at enterprise scale.
Priyanshu Jha (Senior Engineering Manager)
AtlassianPranav Shahi (Vice President Information Technology) Mozilla
Arshdeep Singh (Member of Technical Staff) Thinking Machines Lab
Andy Singh (CEO @ Teqfocus)
Neeharika Kompala (Engineering) Microsoft
Moderator: Jonathan Pasky
Executive Producer & Founder @ DevNetwork
Register here: https://apiworld.co/register/