I started my career at Target as an individual contributor and spent 14 years there. My academic journey includes three master’s degrees—in genetics and statistics, software engineering, and an MBA—which helped shape my holistic thinking. During my time at Target, I worked in multiple domains, which gave me a comprehensive understanding of the retail business and how technology evolves to meet growing business needs.

I began by pulling data from various source systems into data warehousing, then transitioned to the Hadoop ecosystem, working on clickstreams, personalization, advanced machine learning for optimization, and resource allocation. I also led IoT initiatives, like using minute-by-minute data to optimize HVAC systems and forecast electricity demand.

After Target, I joined UnitedHealth Group as a VP and spent five years focusing on personalization, member engagement, and retention. I also led data efforts for NaviHealth, where we developed predictions for length of stay and reduced patient readmission risks. This role involved highly sensitive and empathetic work, using data to improve patient outcomes.

Now, at Coca-Cola, I bring these experiences together to drive business outcomes with data and technology. Key lessons I’ve learned include understanding organizational maturity, culture, and domain while focusing on what’s possible to achieve meaningful impact.

Coca-Cola is one of the largest marketing companies, and balancing innovation, speed to market, and strong fundamentals is critical for us. It’s about maintaining a clear perspective on whether initiatives are short-term experiments or long-term strategies. Leadership plays a key role in finding this balance.

For example, Coca-Cola was among the first to experiment with OpenAI, leveraging creativity, a smart team, and strong technology partnerships to achieve quick wins. However, not everything can be done at that pace. It’s essential to match the right technologies to the right use cases, guided by strategic thinking and governance principles.

There’s no one-size-fits-all solution; technologies like GenAI excel in some areas but aren’t universal fixes. The key is balancing experimentation and innovation with investments in foundational elements, while always keeping our core business in focus. Ultimately, we’re not a tech company—we’re a company that sells Coke—and our data strategy reflects that.

I think it’s very important to keep that balance and also remembering  what is our core business.  and keep that core competencies on the top. Um, we may get excited with technology, but also remembering that, you know, we are not a tech company. We are a company who sells Coke. So those are the balance things that we need to keep and embed into the data strategy.

A successful data strategy involves everyone in the organization, not just the technology or business teams. At Coca-Cola, it’s a business-led, tech-enabled effort focused on achieving actionable outcomes.

Key components include prioritizing value-driven, needle-moving use cases to ensure focused investments, while also building scalable, modular, secure architectures. We are embracing advanced technologies like confidential computing to address the needs of our extensive ecosystem of branches, partners, customers, and consumers.

Another priority is expanding in-house capabilities to ensure solutions are repeatable and adaptable to multiple use cases. This involves integrating diverse data types—first-party, second-party, third-party, and internal data—through robust taxonomies, ontologies, governance frameworks, and a unified golden record.

Talent and culture are also critical. We invest in upskilling teams and fostering collaboration between new and existing talent to create a cohesive, innovative environment. Lastly, we are also investing in automation, observability, and making sure we are tracking to our guidelines for security, privacy, as well as ethics and compliance.

With rapid advancements in technology, such as data observability, data modeling, MLOps, and confidential computing, how do you prioritize short-term needs versus long-term goals when managing data sharing across internal and external parties?

It’s about focusing on needle-moving use cases. For instance, with confidential computing, we don’t approach it solely as a tech initiative. Instead, we collaborate with partners to create win-win scenarios, embedding it into use cases.

This way, we build capabilities while delivering tangible value to the business. While some experimentation happens within tech teams, the primary driver is always the use case and the value it generates.

It’s about focusing on needle-moving use cases. For instance, with confidential computing, we don’t approach it solely as a tech initiative. Instead, we collaborate with partners to create win-win scenarios, embedding it into use cases.

This way, we build capabilities while delivering tangible value to the business. While some experimentation happens within tech teams, the primary driver is always the use case and the value it generates.

Every organization has a different approach. Navigating cloud adoption depends on an organization’s unique business needs and strategy. Companies like Target, UnitedHealth, and Coca-Cola have taken different approaches, reflecting their cloud maturity levels. Some use hybrid or multicloud options, while others stick to a single platform.

A key fundamental is ensuring robust data governance for consistent, reliable reporting—especially for performance metrics. This includes maintaining one source of truth and a well-defined glossary. While this may seem traditional, it’s still critical for effective business measurement.

On the other hand, advanced use cases, such as product-centric applications, require intelligent data products that drive personalization, optimization, and demand forecasting. These need to be integrated into operational systems, supported by strong architecture and MLOps or LLMops.

Successful cloud adoption also requires flexibility—whether using batch or real-time architectures like Lambda, depending on business needs. For instance, real-time is crucial for companies like Uber. The key is balancing standardization for critical business functions with the agility to rapidly build and deploy products. This combination has proven effective.

We’re seeing a platform approach to GenAI deliver better returns for many CPGs.

Additionally, organizations still struggle with “death by dashboard,” where too much focus is on chasing answers rather than asking the right questions. Do you think GenAI can alleviate this issue?

We’ve learned a lot over the past year or two about what GenAI can really do and where we should focus its capabilities. It’s not a silver bullet—there’s no magic with one technology. Dashboards, for instance, are static, and when people look at them, they have more questions. The key is answering those questions with accurate data, and that’s where GenAI can help, as long as we’re careful.

A lesson we all learned is hallucinations. The data we feed GenAI is critical. It needs to be accurate and consistent to get the right responses. This is where some technologies, like knowledge graphs, can help. By embedding them into the data mix, they can contextualize and infer, reducing hallucinations and improving the reliability of the answers GenAI provides.

At Coca-Cola, we’re using this combination of technologies—knowledge graphs layered with large language models (LLMs)—to build trust in the responses we get from GenAI assistants or agents. That’s a challenge and is also where the experience comes into the mix. But it’s also about the questions we ask. We’ve found that how you write your prompts matters, and helping users craft better prompts is part of the ongoing learning process.

I would also mention the due diligence we need to make. It’s similar to how we would treat information elsewhere—would you trust a peer-reviewed article over something you read on a wiki? It’s the same principle: we need to ensure the data we feed into GenAI is trustworthy, and we need to be diligent in interpreting the answers it provides.

Another important aspect is ethics, privacy, and security. What we feed into GenAI comes back to us, so we need to ensure what we are feeding to the engine is good.

Be curious – keep your feelers on – keep learning – be bold – take challenges. And, celebrate wins and failures equally, and that’s only going to help you to do even better.

I will repeat the same things if I had to do it all over again.