Twenty years ago, I was staring at a spreadsheet of consumer payment histories at Citi, and I had no idea what it was trying to tell me. Rows of numbers, no story. It took time, mentorship, and a lot of trial and error before I understood the gap between having data and having insight, and closing that gap has, in one form or another, been my job ever since.

I’m Charles Mulwa. Over the past two decades I’ve built and led data teams across four industries that have almost nothing in common on the surface (banking, healthcare, tax, and now aviation), and I currently serve as Manager of Data Management and Governance at Southwest Airlines. People sometimes assume that kind of industry-hopping means starting over each time. In my experience, it’s the opposite. The tools change constantly. What doesn’t change is the discipline underneath: earn the business’s trust in your data, then protect that trust as the organization scales.

Here’s how that discipline took shape, one industry at a time.

Citi: Learning That Data Has Consequences

My career started in business intelligence at Citi, where the mandate was blunt: use data to reduce financial losses. That meant digging into consumer behavior and payment patterns to understand who was likely to default, and why, before it happened.

There was no room for “close enough” in that work. A risk model isn’t an academic exercise. It’s a decision that affects real accounts and real people, and it shows up on the bank’s balance sheet whether it’s right or wrong. That was the first time I understood, viscerally, that data work carries consequences. It’s a lesson I’ve never had to relearn, only reapply.

Fourteen Years in Healthcare: Phytel and IBM Watson Health

In 2008 I moved into healthcare, first at Phytel and later at IBM Watson Health after IBM’s acquisition of the company. I ended up staying in this space for fourteen years, and it shaped my thinking about data more than any other stretch of my career.

At Phytel, the goal was proactive care: curating and delivering data so providers could catch problems early instead of reacting after the fact. That mission scaled significantly once I moved into data management at IBM Watson Health, where I led efforts to integrate demographic, clinical, and encounter data in service of population health, helping identify care gaps and connect patients to the right providers before those gaps widened.

Healthcare taught me something financial services hadn’t: the cost of bad data isn’t always a bad decision on a spreadsheet. Sometimes it’s a patient who falls through the cracks. That’s when data governance stopped being an abstract best practice for me and became something closer to an obligation. Who owns this record? Who’s accountable when it’s wrong? What happens to the person on the other end of it if we get it wrong?

Fourteen years is a long run in one industry. I don’t regret any of it, but eventually I wanted to find out whether the governance instincts I’d built in healthcare would actually hold up somewhere completely different.

A Shorter Chapter With a Lasting Lesson: Ryan LLC

In 2022, I stepped into a role as Director of Data Engineering at Ryan LLC, a firm built around corporate tax services. The mission here was more concrete than almost anything I’d worked on before: use data to help clients avoid overpaying in taxes, freeing up capital they could put back into their business.

It didn’t last long, but it left a mark. Tax and audit work runs on defensibility: every figure has to be able to withstand scrutiny, because the risk isn’t just financial, it’s regulatory. That reinforced something I already suspected but hadn’t fully tested. Data engineering isn’t just pipelines and infrastructure. It’s building systems that produce numbers you can stand behind when someone challenges them.

It also sharpened a perspective I carry into every role now: data has to serve the client’s actual outcome, not just check an internal reporting box.

Southwest Airlines: Where Everything Converges

Since August 2023, I’ve led data management and governance work at Southwest Airlines. In a lot of ways, this role is where the previous seventeen years converge.

Aviation is its own kind of high-stakes environment, but the underlying truth is familiar from banking and healthcare alike: operational efficiency, cost control, customer experience, and regulatory compliance all depend on how well an organization manages its data assets. When data is fragmented or untrustworthy, it doesn’t stay an abstract IT problem. It turns into friction that employees and customers both feel, often in real time.

What keeps this role interesting for me is the sheer scale of it. This isn’t governance in theory; it’s making sure that millions of daily operational decisions, across a massive enterprise, are backed by data that’s accurate, accessible, and properly governed. Twenty years in, that kind of challenge is exactly what keeps this work feeling far from routine.

Five Lessons the Whole Journey Taught Me

Step back from the individual companies, and a handful of lessons keep resurfacing, the ones I now bring into every team I lead and every strategy conversation I’m part of.

Governance is a people problem wearing a technology costume. I’ve built governance frameworks in banking, healthcare, tax, and aviation, and the technology has never been the hard part. The hard part is getting a large group of people to agree on ownership, definitions, and accountability. The best data catalog in the world won’t save a governance program that nobody trusts.

Build for the organization you’ll be, not the one you are. Early in my career, I watched too many data solutions get built for the problem directly in front of them, with no thought for what happens when the company doubles or the regulations shift. I’ve made it a habit ever since to design with room to grow, not just room to function today.

Think like an owner, not a technician. Somewhere along the way, my decision-making shifted from “is this technically correct” to “what does an owner of this business need from this data,” weighing profitability, compliance, and stakeholder impact together instead of optimizing for just one. It’s probably the single biggest change in how I lead compared to when I started.

The people you mentor outlast the systems you build. Some of the most meaningful parts of my career haven’t been the platforms or pipelines. They’ve been the analysts and engineers I’ve mentored who’ve since gone on to lead their own teams. I don’t treat that as a soft skill layered on top of data leadership. In my experience, it’s inseparable from it.

Industry expertise travels further than people expect. I’ve moved from financial risk to population health to tax strategy to airline operations, and each move felt, on the surface, like starting over. Underneath, though, the core discipline (building trustworthy data foundations that people can actually act on) came with me every time. If you’re early in your career and worried a change of industry means losing your expertise, my experience says otherwise.

Say the hard thing early. Every governance failure I’ve seen up close had an earlier moment where someone suspected the data was wrong and didn’t say so, because it wasn’t their job, or because the meeting had already moved on, or because raising it felt like slowing the team down. The leaders I respect most are the ones who build a culture where flagging a bad number is rewarded, not treated as an inconvenience. I try to be that kind of leader, and I don’t always succeed, but I’ve gotten better at it with every industry I’ve worked in.

The Tools Keep Changing. The Discipline Doesn’t.

It’s worth naming just how different the technology looks today compared to when I started. In 2006, “the data warehouse” was the center of the universe: a single, tightly governed structure every report pulled from. By my years at IBM Watson Health, the conversation had moved to data lakes, cloud migration, and the early promise of machine learning applied to population health. Today at Southwest, it’s cloud-native platforms, real-time pipelines, and AI-assisted analytics, layered on top of the exact same governance fundamentals I learned two decades ago.

I’ve written elsewhere about why I think the traditional data warehouse isn’t dead so much as it’s evolving, and living through that evolution firsthand, rather than reading about it after the fact, is a big part of why I hold that view. My advice to younger data professionals I mentor is simple: enjoy the tools, chase the new platforms if that excites you, but don’t mistake tool fluency for data leadership. The professionals who last twenty years in this field are the ones who understand that platforms come and go, and governance discipline doesn’t.

I’d add one more thing to that advice: don’t wait for a title to start thinking like a governance leader. Some of the strongest instincts I see in early-career analysts have nothing to do with their job description. They’re the ones who ask “where did this number actually come from” before they build on top of it. That habit is more valuable, long-term, than fluency in any single platform, and it’s the one thing I look for consistently when I’m hiring or mentoring someone new to the field.

The Common Thread Underneath Four Very Different Jobs

People sometimes ask what it’s like to move between industries as different as banking, healthcare, tax, and aviation, as if each shift meant relearning the job from the ground up. In some ways it did: the regulations are different, the stakeholders are different, and “a good outcome” means something different in each one. A good outcome at Citi was a more accurate risk score. A good outcome at IBM Watson Health was a patient connected to care before a condition got worse. A good outcome at Ryan LLC was a client legally keeping capital that was theirs to keep. A good outcome at Southwest is an operation that runs smoothly because the data behind it held up.

But strip away the industry-specific language, and it’s always the same job: earn the business’s trust by proving the data is right, then build the governance and infrastructure to keep it right as things scale. Every organization I’ve worked for has eventually asked some version of the same question: can we trust this number enough to act on it? A data leader’s real job is being able to answer yes, honestly, every time.

Why I Still Show Up for This Work

I don’t think twenty years qualifies anyone to say they’ve fully figured this out. If anything, the questions have only gotten more interesting as AI reshapes what “governed data” even means in practice. How you govern a model’s inputs and outputs is a different challenge than governing a static report, and it’s one the field is still working out in real time.

What hasn’t changed is what pulled me into this work in the first place: the moment when a dataset stops being rows and columns and starts telling you something true about the business, or the patient, or the operation behind it. Chasing that moment is what’s kept me in this field across four industries and twenty years, and it’s what I expect will keep me here for a while longer.

I write here to share what I’m still learning about data leadership, governance, and building teams that last, not because I have it all figured out, but because I think this field benefits when people are honest about what actually works. If you’d like a slightly different take on this same journey, I also write at charlesmulwa.blogspot.com, where I go into a few of these moments in more depth. If that’s useful to you, I’d love to stay connected. Subscribe below or find me on LinkedIn.