With over 170 years of history, American Express has evolved into one of the world’s most recognized financial services brands, serving over 140 million customers across the globe. While the company is known for its premium card offerings and global merchant network, it is also at the forefront of enterprise data innovation, investing heavily in AI, cloud transformation, and customer-centric intelligence.

This two-part interview series features Purvi Shah, VP, Enterprise Data Platforms at American Express, in a candid conversation with Rohit Choudhary, CEO of Acceldata. Shah, who has spent 15 years at American Express, offers a behind-the-scenes look into how the company is building its next-generation cloud-native data platform, Lumi, to power everything from fraud detection to real-time personalized offers.

She also reflects on her career journey, the critical role of data governance, and how Amex is preparing its workforce for a fast-moving AI future.

Edited Excerpts

Q: To begin, could you walk us through your role at American Express? What does your team focus on, and how are you helping drive enterprise data forward?

I’ve been at American Express for 15 years, and I currently lead the enterprise data products team. Broadly, my role covers three key areas. First, I lead our enterprise data platforms, which includes our big data ecosystems — Cornerstone, our on-prem platform, and Lumi, our next-generation big data platform on the cloud.

Second, I oversee our business intelligence suite of products.

And third, I manage our enterprise data management tools, including data quality as a service, data retention and deletion solutions, and our enterprise data catalog.

Q: Can you tell us a bit about your career journey, how you got started and what led you to your current role?

I did my undergrad at Penn State University in Information Science and Technology, and that’s where my passion for data began.

My first job involved going through reams of data from Netscape. I analyzed user search queries to identify patterns and help optimize search performance on their website. That was my first experience with the power and complexity of data.

After that, I worked at a consulting firm supporting public sector clients. I implemented large-scale technology solutions, like food stamps and early childcare intervention programs.

Later, I earned my graduate degree from the University of Texas at Austin (McCombs School of Business) and joined American Express in 2010. Since then, I’ve worked across several teams — technology, merchants, and customers. I’ve led products like American Express Offers, which creates value for both cardholders and merchants by delivering personalized deals.

Q: Looking ahead, how do you see the future of data management evolving? And what skills do you think leaders need to keep up, especially in context of your Lumi platform vision?

Change is happening fast. What I remind myself and my team is: we have to stay curious and keep learning. That learning can come in many forms — reading the news, exploring LinkedIn Learning, or even having interviews like this where we learn from each other’s experiences.

There are lessons and pitfalls across the industry we can learn from. At American Express, we’re very focused on upskilling our workforce.

Specifically for Lumi, our next-gen public cloud-based big data platform. It’s a brand-new asset that will help us deliver on our customer promise. We’ve already trained over 15,000 internal colleagues to use the new technology and product capabilities that Lumi enables.

Q: With American Express’s massive global customer base, how are you leveraging AI and machine learning to enhance customer experiences, especially as technologies like GenAI evolve?

Data is at the heart of everything we do, and technology enables that. We’ve been on a long journey of modernization focused on enhancing the customer experience.

Even 15–20 years ago, we were early to embrace big data trends. Our on-prem platform, Cornerstone, enabled us to build traditional AI models — for example, for fraud detection. We’ve developed some of the best fraud detection capabilities in the industry because we can bring data together, derive insights fast, and deploy AI models at scale.

Now, as we transition to Lumi and the public cloud, we can use more advanced models, including GenAI. These allow us to move even faster, understand customer intent better, and deliver real-time, relevant offers.

For example, if a customer is looking to upgrade their card, we can understand that intent and offer something that’s timely and valuable.

Offers are a key part of how we bring value to both customers and merchants. With Lumi and AI, we’re able to provide a seamless omnichannel experience, whether someone’s on our app or website, making sure the right offer reaches the right person at the right time.

Q: Last time we spoke, you shared an analogy about running an apartment complex to explain Lumi’s role. Can you walk us through that again, especially as more teams begin accessing data and building models?

One of the great things about American Express is how seriously we’ve always taken data — organizing it in the right way, using the right technology stack to derive insights, and making faster decisions to power great customer experiences.

The apartment analogy helps me explain this even to my parents. I describe Lumi as the apartment building that I own. In this building, I provide shared services like data movement. Think of that like an elevator that takes you from the ground floor to the penthouse.

Then there are the tenants — our internal data owners — who bring their data into the building. That’s where data governance comes in. Just like you wouldn’t let a tenant in who can’t pay rent or isn’t credible, we make sure we understand the data, the metadata, where it can be used, who has access, and what privacy protections are required.

The apartment analogy clarifies where my responsibilities begin and end, and how shared services, like one central elevator instead of ten, can benefit everyone. It’s about common patterns, reusable services, and driving efficiency, consistency, and speed across our 15,000+ data users.

Q: From an AI-readiness perspective, how are you thinking about data quality and governance in Lumi? Is there a broader conversation needed within the organization around preparing data for next-gen use cases?

All of us across the industry are learning about it. At American Express, we’ve implemented additional processes to make sure GenAI use cases are piloted responsibly. We’re focused on avoiding customer harm (intentional or unintentional) and making sure the outcomes align with customer needs.

It’s not just about deploying an LLM. We use both our internal models and commercially available ones. That combination makes us stronger and we don’t want to reinvent the wheel, we want to accelerate.

The key is managing the full process, with data at the center. We want to avoid “garbage in, garbage out,” so we focus not only on traditional data quality but also on accuracy, recency, and proper permissions for every use case.

Controls are critical, and we’re constantly strengthening them. We’re also working to align Lumi with our traditional modeling platforms so we can scale responsibly and efficiently.

CDO Magazine appreciates Purvi Shah for sharing her insights with our global community.