By: Zach Miller

Srinivasarao Paleti, a banking and telecom expert, has spent over 15 years in banking and compliance, rising through the ranks at Tata Consultancy Services. His path led him from telecom systems to AI research, where he now focuses on using machine learning to make financial systems smarter and safer.

In this interview, Srinivasarao discusses how he reshapes the use of technology. With over 15 published papers, two patents, and a special focus on fraud prevention, Srinivasarao is helping redefine financial trust in the digital age. He also discusses the risks banks face, the role of agentic AI in solving them, and why real-time decision-making matters more than ever. He explains how AI works, why it matters, what it solves, and where it’s headed next.

You have deep experience across both telecom and banking sectors—how has this dual exposure shaped your approach to building AI models that are both agile and compliant?

My experience in both telecom and banking has given me a unique perspective on AI system design. Telecom taught me scale and real-time processing, whereas banking emphasized compliance, risk mitigation, and trust. Combining these insights, I focus on building AI models that are not only robust and agile but also grounded in auditable, regulatory-aligned architectures. This cross-domain exposure enables me to create systems that can evolve quickly without compromising on security or governance.

Many financial institutions struggle to balance innovation with regulation. In your experience, how can AI systems be built to remain adaptive without violating strict compliance frameworks?

The key lies in architecting AI systems with compliance as a foundational design principle, not an afterthought. This means embedding traceability, explainability, and human oversight from the start. By adopting modular AI components, institutions can upgrade or replace parts of the system without disrupting the entire compliance framework. Additionally, using sandbox environments and governance frameworks allows experimentation and innovation while maintaining regulatory guardrails.

You’ve authored books, published research, and hold patents in AI—how do you translate academic innovation into enterprise-ready solutions within legacy banking systems?

Bridging academic innovation with enterprise application starts with abstraction—distilling complex algorithms into modular components that can integrate with existing systems. I focus on outcome-driven implementation, where theoretical advancements are mapped to tangible business use cases like credit scoring or fraud detection. Legacy banking systems require reliability, so we encapsulate novel AI within proven engineering patterns—ensuring innovations are not only effective but also production-grade, resilient, and regulatory-compliant.

Agentic AI is a recurring concept in your work. Can you explain what that means in a practical sense, and how it improves on rule-based automation in areas like KYC or AML?

Agentic AI refers to AI systems that possess goal-directed behavior and autonomy, making decisions based on context rather than predefined rules. In practical terms, for domains like KYC or AML, this means the AI can dynamically assess risk, learn from evolving fraud patterns, and adjust its actions in real-time. Unlike rule-based systems that require manual updates for new threats, agentic AI adapts proactively—enhancing detection capabilities while reducing false positives and operational overhead.

Auditability and explainability are often treated as afterthoughts in AI. How do you ensure these principles are embedded from day one of development?

Auditability and explainability are engineered into the system architecture from the outset. I employ model governance frameworks that log every decision pathway, including data lineage, model parameters, and decision outcomes. Additionally, we integrate interpretable AI techniques—like SHAP or LIME—within the pipeline to provide human-understandable justifications. This ensures that compliance teams, auditors, and business users can trust and validate AI behavior, even in complex decisioning scenarios.

Looking 3–5 years ahead, what key risks do you foresee in financial AI systems if institutions do not adapt to emerging technologies or regulatory shifts in time?

If institutions fail to adapt, the most pressing risks include systemic bias, model drift, and regulatory non-compliance. As AI models become more complex, black-box systems may introduce opaque risks that are difficult to trace or mitigate. Moreover, cyber threats targeting AI pipelines could exploit vulnerabilities in model training or data ingestion. Without proactive governance and continuous adaptation to new technologies and regulations, institutions may face financial penalties, reputational damage, or even systemic failures.

Conclusion

Financial systems move faster than ever, and trust can’t be an afterthought. Srinivasarao Paleti knows this better than most. He is making sure that as banks embrace AI, they don’t lose sight of transparency and responsibility.

From adaptive KYC to fraud detection powered by agentic AI, Srinivasarao is pushing boundaries but keeping both feet on the ground. He reminds us that AI in finance is not just about speed or scale; it is also about precision and accuracy. It’s about building systems that respond, explain, and improve with time. Srinivasarao wisely advises: Stay curious, keep your systems explainable, and never lose sight of the people behind the numbers. After all, trust is still the most valuable currency.

Disclaimer: The views and opinions expressed in this interview are those of the interviewee and do not necessarily reflect the official policy or position of any organization or entity. The information provided is for general informational purposes only and should not be construed as legal, financial, or technical advice.