As Artificial Intelligence moves from experimentation into the backbone of critical infrastructure and high-stakes decision systems, the key question is no longer how powerful models are, but how much we can trust them.

This two-part session explores the end-to-end journey of building trustworthy AI systems in real-world environments. The first talk examines how AI is democratizing the creation of sophisticated statistical models using structured infrastructure data to assess change risk in cloud and platform systems. The second talk shifts focus to model calibration—an often-overlooked but essential discipline for ensuring AI predictions remain reliable, interpretable, and decision-ready in regulated domains such as finance and healthcare.

Together, these talks offer a practical roadmap for engineers, data scientists, and AI practitioners to design, evaluate, and deploy AI systems that not only perform well—but behave responsibly under uncertainty.

Session 1: AI-Powered Statistical Models for Infrastructure Change Risk Assessment

Time: 5th Feb, 3:00 PM – 3:30 PM EST

Speaker: Dylan Ratcliffe, Founder & CEO of Overmind

Speaker Bio

Dylan Ratcliffe is the Founder & CEO of Overmind, bringing deep expertise in infrastructure automation and tooling. Prior to founding Overmind, Dylan spent 6 years at Puppet, where he led their international Professional Services business. Recently relocated from London to San Francisco, Dylan combines hands-on infrastructure experience with entrepreneurial vision to solve complex problems in the DevOps and cloud infrastructure space.

Description: In this talk, Dylan Ratcliffe demonstrates how modern AI tools are democratizing the creation of sophisticated statistical models—work that traditionally required dedicated data scientists. Using Terraform as a primary source of structured infrastructure data, Dylan will walk through the development of custom mathematical models designed to distinguish "routine" infrastructure changes from those that pose a high risk to system stability.

While the production system is built in Go, this session focuses on the methodology and mathematics behind the models. Attendees will gain insights into:

• How AI accelerates the development of domain-specific statistical models.

• Leveraging Infrastructure-as-Code (IaC) data for intelligent risk analysis.

• Bridge the gap between raw infrastructure data and actionable risk assessments.

Session 2: Model Calibration: The Hidden Key to Trustworthy AI in Finance and Healthcare

Time: 5th Feb, 3:30 PM – 4:00 PM EST

Speaker: Swati Tyagi, AI/ML Leader & Researcher

Speaker Bio

Swati Tyagi is an AI/ML leader and researcher specializing in responsible AI, generative AI, and data-driven decision systems for highly regulated industries. With a PhD in Statistics and extensive industry experience, she has led impactful work in bias mitigation, model evaluation, and large-scale AI deployment. Swati is an active speaker, mentor, and community builder, contributing to global tech forums, academic research, and professional communities to advance ethical and trustworthy AI.

Description: In high-stakes domains like finance and healthcare, accuracy alone is insufficient; models must also be calibrated to reflect true uncertainty. A model that is 90% "confident" but only 60% "accurate" can lead to catastrophic failures in credit risk or clinical settings.

Dr. Swati Tyagi explains why highly accurate models can still cause harm when probability estimates are unreliable. Using real-world examples from fraud detection and medical decision support, this session covers:

• Diagnosing Miscalibration: How to tell when your model is "overconfident."

• Practical Techniques: Methods to align predicted probabilities with actual outcomes.

• Production Monitoring: Strategies to ensure AI systems remain trustworthy and decision-ready over time.