IEEE Women in Engineering Webinar on Trustworthy AI Systems
Join IEEE Women in Engineering for an insightful webinar on building trustworthy AI systems—a critical challenge as AI moves from experimentation to real-world, high-stakes applications.
This session brings together two experienced speakers who will share practical perspectives on why many AI systems fail to gain trust and how we can design systems that are reliable, safe, and transparent by design.
Through real-world case studies and research-backed frameworks, you’ll explore how to build AI systems that not only perform well but also know their limits, communicate uncertainty, and prevent critical failures. Topics will include production challenges in AI deployment, reliability and explainability, and safety mechanisms for high-impact environments such as healthcare.
Whether you’re a student, engineer, or industry professional, this webinar will equip you with actionable insights, practical frameworks, and key questions to evaluate and build trustworthy AI systems.
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
Loading virtual attendance info...
- Starts 27 April 2026 04:00 AM UTC
- Ends 04 June 2026 04:00 AM UTC
- No Admission Charge
Speakers
Anusha Kovi of Amazon
From Demo to Durable: How to Ship AI Systems People Actually Trust
88% of AI pilots never reach production, not because the AI doesn't work, but because people don't trust the output. In this talk, I'll walk through a real-world case study of an AI analytics system that failed in production and the practical architecture we built to fix it.
You'll leave with a concrete five-layer framework, covering policy, retrieval, reasoning, verification, and explanation, that any engineering team can apply to build AI systems that are both accurate and trustworthy. This talk provides the questions to ask and the red flags to watch for when building or evaluating AI tools.
Biography:
Anusha is a Business Intelligence Engineer at Amazon with 6+ years of experience in Data and AI. She specializes in building production-grade AI systems that are reliable, explainable, and safe. A frequent speaker at technical conferences, she is passionate about making AI trustworthy by design.
Latha Iyer
Building Safety Alerts for High Stakes Systems
Artificial intelligence is being deployed in hospitals, clinics, and life-critical settings faster than we have built the tools to govern it. The most dangerous failure is not when AI gets something wrong. It is when AI gets something wrong with confidence.
This talk introduces the Adaptive Confidence Circuit Breaker (ACCB) — a framework that monitors every AI prediction in real time and raises an alert the moment the system cannot be trusted. Like an electrical circuit breaker that trips before damage is done, ACCB stops an unreliable output before it reaches a doctor, a patient, or a decision-maker.
Validated on real ALS patient biomarker data (Lu et al. 2015, 219 patients) and mitochondrial intervention modelling, ACCB demonstrates that it is possible to build AI systems that know the boundary of their own reliability — and say so.
Biography:
Latha Iyer is the designer of the Adaptive Confidence Circuit Breaker (ACCB), a cross-domain AI safety framework validated on ALS biomarker data and mitochondrial intervention modelling. She holds an M.S. in Business Analytics from the University of Louisville (GPA 3.926) and brings 24 years of enterprise AI delivery experience across healthcare and financial services. Her research is published on Zenodo with an arXiv submission in progress.
Agenda
Session 1 (45 minutes)
Talk: From Demo to Durable: How to Ship AI Systems People Actually Trust
Speaker: Anusha, Business Intelligence Engineer, Amazon
- Presentation (30 minutes)
- Why most AI pilots fail in production
- Real-world case study of an AI system failure
- A practical five-layer framework: policy, retrieval, reasoning, verification, and explanation
- Key questions and red flags when building trustworthy AI systems
- Q&A (15 minutes)
Session 2 (45 minutes)
Talk: When AI Says “I Don’t Know”: Building Safety Alerts for High-Stakes Systems
Speaker: Latha Iyer, Applied AI Researcher
- Presentation (30 minutes)
- Risks of overconfident AI in critical environments
- Introduction to Adaptive Confidence Circuit Breaker (ACCB)
- Real-world validation using healthcare data
- Designing AI systems that recognize and communicate uncertainty
- Q&A (15 minutes)