Engineering Trust into AI Systems: Designing for When Things Go Wrong

Production AI systems will fail, encounter unexpected inputs, and operate under conditions their designers cannot fully anticipate. In “Engineering Trust Into AI Systems: Designing for When Things Go Wrong,” Cal Al-Dhubaib of Rubrik examines a critical enterprise AI challenge: building trust by designing for failure, resilience, and real-world operational risk from the start. As organizations move AI from pilots to production, reliability becomes a systems problem—not simply a question of model performance. Enterprise teams need to think about what happens when AI systems behave unexpectedly and how architecture, engineering practices, governance, and operational controls can support trustworthy AI at scale. This session brings a cloud and AI technology perspective to the practical work of engineering production-ready AI systems. It connects directly to the broader enterprise need for AI systems that are not only capable, but also resilient, governed, observable, and designed to operate under real-world conditions. KEY TAKEAWAYS • Explore why trustworthy AI requires designing for situations where systems do not behave as expected • Consider reliability and resilience as core requirements for production AI systems • Examine the relationship between AI engineering, operational controls, and enterprise trust • Understand why moving from AI pilots to production requires a broader systems perspective • Gain an enterprise technology perspective on building AI systems for real-world operating conditions SPEAKER 📢 Cal Al-Dhubaib, Principal Technologist, Cloud and AI | Rubrik ABOUT AI ENTERPRISE CONFERENCE 2026 Hosted by Data Science Connect on September 1, 2026 in New York City, AI Enterprise Conference 2026 brings together Fortune 500 decision-makers, enterprise data and AI leaders, and leading technology providers to explore how organizations can build, scale, govern, and operationalize AI. The conference theme, “AI for Enterprise: From Pilots to Production,” focuses on the systems, governance, data foundations, economics, and operating models required to move enterprise AI into scalable, secure, production-ready environments tied to real business outcomes. Watch the full session to explore what it takes to engineer greater trust and resilience into enterprise AI systems. Subscribe to the Data Science Connect YouTube channel, explore more AI Enterprise Conference 2026 sessions, and discover additional insights from enterprise data and AI leaders.