Beyond test coverage: engineering trust in AI-powered systems
-
Locked
QA lead
Talk Description
AI systems don't fail the way traditional software fails and your test suite wasn't built for the difference.
In this session, Tanvi Mittal, Senior SDET and Test Automation Lead at U.S. Bancorp, makes the case that test coverage is a necessary condition for shipping AI, but not a sufficient condition for trusting it.
Drawing from real fintech scenarios and three original frameworks the SAFE-R rubric, PACE, and AGENT-F taxonomy Tanvi shows how trust in AI systems is measurable across five concrete dimensions: accuracy, consistency, safety, fairness, and explainability. Attendees will leave with a practical threat model for agent failures, an understanding of how regulatory requirements like SR 11-7 and Reg E map directly to AI test design, and three actions they can apply to their own systems the following week.
Tanvi Mittal is an AI quality engineer and test automation leader specializing in testing LLM and agentic systems. She is a WCSC 2026 keynote speaker, IEEE Senior Member, and community mentor.
Tanvi Mittal is an AI quality engineer and test automation leader specializing in testing LLM and agentic systems. She is a WCSC 2026 keynote speaker, IEEE Senior Member, and community mentor.
