Hello everyone. I'm Ashish Lia, managing director and principal Data architect at Webster Bank, where I lead large scale data and AI initiatives across regulatory reporting, enterprise data platform, and intelligent financial system. I'm delighted to be here to share insight on AI data governance for scalable compliance in modern banking. At con 42 Cube native 2025 conference. So AI is no longer an experiment in banking. It's a mission critical from fraud detection to risk assessment to customer engagement. AI is enabling in every layer of the modern financial institute. But with this power comes a set of governance challenges that we must address quick glimpse of the agenda. We will begin with the context and introduction, why modern governance has become imperative. Then I'll walk you through the evaluation of ai. Next we'll mine the limitation of the traditional governance model. From there, I will introduce to unified AI data governance framework. And we'll explore compliance and regulatory alignment, followed by a real world case studies. And then we'll close with the future roadmap and some key take of this. So first part of the session, introduction and context. Banking operates in one of the most banking operations is one of the most regulated environment in the world today. Compliance failure don't just bring financial penalties. Sometimes it is billions. They also damage the reputation. And at the same time, bank face intense pressure to innovate rapidly deploying AI to meet the customer and market demands traditional framework designed for static data warehouse. And predictable. ETL simply cannot handle the complexity and the dynami dynamism of AI system. So this dual pressure one, the compliance on one side. And then innovation on the other side make it essential to rethink the whole governance process. Jumping onto the next segment of of our discussion that is AI evaluation and containerization in the banking. So over the last decade AI in banking has grown from a small, isolated pillar. Into an enterprise scale system that basically enables a core banking operations. Think about a fraud detection model. They take input from dozens of data sources, analyze them in milliseconds, and continue to learn in the real time. So AI is no longer a tool, it is a backbone of the modern banking infrastructure nowadays. Talking about the containerization revolution to support this scale bank has embraced the centralization or containerization. Kubernetes enable AI workload to run with incredible flexibility deploying across the hybrid cloud scaling dynamically and updating seamlessly through blue screen or cannery deployments. But here is the trade off. While the containers bring agility, they also bring complexity. The complexity of workload, dynamic service meshes distributed clusters, traditional monitoring and governance tool struggle to keep up with all of this. And that's where the governance challenge become more prominent or deepens basically. Quickly talking about the challenges with the traditional governance model. Traditional governance model falls short in four key ways. If you just broadly look at these four segments, it is going to cover the challenges with the traditional governance model. The first one is the metadata management. So AI produces massive amount of constant changing metadata. Then the second is lineage tracking. Understanding where each feature of the model input comes from is nearly impossible at a scale. Third, real time compliance. So regulators demand continuous monitoring of biased drift and quality. And fourth is regulatory comp complexity. Laws now demand explainable fair and bias-free ai. These are not just check boxes, they are the continuous obligation. As of now, just moving on to the next segment of discussion that is unified AI data governance framework. So this is where new approach is required. The unified AI data governance framework is built for cloud native containerized infrastructure. It deploys governance as microservice that is scaled dynamically, deeply integrated with the Kubernetes orchestration. And it uses a seamless metadata architecture, which enable the capturing and analysis of events in real time. Ensuring governance isn't lagging behind ai but operates alongside it. The framework has three critical components. The first one is the metadata orchestration engine. The central nervous system collecting the basically the managing collection of managing thousands of metadata events per second. A second part is the intelligent lineage tracker using the graph algorithm and instrumentation to trace dependencies across our distributed AI system. Third one, the real time. Compliance monitor monitoring constantly evolving AI decision against internal politics and regulatory requirement. Together these components make governance containers automated and intelligent. Quickly jumping on to the next, topic, which is to make this work. Metadata management itself must be high performance. We use distributed storage. So relational and graph databases optimizes optimize for different queries. Frequently access data is kept in memory for sub million second response time. Then a specialized index support, time-based graph and full text queries. And obviously metadata federation enable queries across multiple source without centralizing everything, reduces reducing the cost and latency. Tendency Automation is crucial. Draft the drift detection monitoring monitors for shifts in the data concept and performance biased detection ensures basically the fairness across demographic demographic groups. And it is, it protects attributes. And anomaly detection powered by unsupervised machine learning spot spots, unusual behavior in model of data. Or maybe in other words, governance is no longer passive. It is active, intelligent, and evolving alongside ai. Security is built in from ground up. In this case, we apply defense in depth in depth integrating with the enterprise. IAM system. Access isn't just role based. It is attribute based allowing for the fine grained permission. Every governance operation is no longer with full context and all metadata is protected through end-to-end encryption with support for field level encryption and tokenization. This result compliance and security by design not as afterthought. Jumping onto the next segment of the discussion, which is compliance and regulatory alignment, which is a crucial part of the whole equation. The framework directly aligns with the major regulations for GDPR. It supports consent tracking right management and. The right to be forgotten. Identifying every model trained on a person's data for CCPA it automates consumer right consumer rights require and data sharing, tracking for basal three, it supports risk modeling, validation, tracking, and stress testing. And it anticipates the EU AI Act, which is classifying the AI system by risk label and applying the right governance controls. This isn't reactive com compliance. It is in a nutshell all proactive compliance enforcement or alignment. Just quickly looking at some of the cases, studies, roadmap, and maybe the conclusion. So let's look at some real world example. Global tier one bank rolled. This out in phases, beginning with the fraud detection, credit risk, and regulatory reporting, the results were transformative. 68% faster time to market for new AI model, 70%, 75% reduction in the audit preparation time through automated reporting and 42% fewer risk events. And thanks to early detection. This proves that governance does not slow innovation. It actually accelerates it and keeps everybody in the organization safe looking at the future roadmap and some of the emerging technologies, which kind of addresses some of the need when it comes to data governance. So the roadmap looks ahead to emerging technology. Serverless compliance monitoring functions, the trigger in the real time as even dockers centralization or containerization, policy enforcement. Basically deploy directly. So policies are deployed directly alongside workload, then advance explainability human readable inside in AI decisions. Coming next is a federated governance, which is critical for multi-cloud, multi organization ecosystem. And again, this isn't just about today's compliance. It is in a nutshell about preparing for tomorrow's AI system or AI ecosystem as well. To wrap up, AI governance is both necessity and competitive advantage. The unified AI governance framework shows that bank can stay compliant, secure, and resilient, while still innovating rapidly. But success requires more than technology. It needs organization commitment, investment in skills and alignment between governance and business strategies bank that embraces this balance will lead in the age of ai. With that, I'll wrap up my session. Thank you for your time and attention. And I'll be happy to take any question which might be appropriate for the discussion. Thank you so much. I.