This edition of Coffee with Calyptus features Saba El-Hilo, a seasoned technology executive whose career spans image processing, cybersecurity, mapping infrastructure, fintech, and now identity verification. Saba shares how a relentless pull toward high-stakes, high-complexity problems has been the common thread across every industry she's worked in. From scaling decision-making at Mapbox to designing adaptive fraud systems in an AI-driven world, her insights cut right to the heart of what it means to build technology that people depend on, often without realising it.
Your career path has taken you through a wide range of companies, from image processing research to cybersecurity, to mapping infrastructure, fintech, and now background verification, across engineering individual contributor, data platform leadership, and executive roles. What has been the common thread that's guided your choices, and were there moments where that thread felt unclear or hard to follow?
The common thread has always been a pull toward complexity with real-world impact. I’ve gravitated toward domains where technology sits at the center of high-stakes decisions. Whether that was image processing, cybersecurity, mapping infrastructure, fintech, or now identity and trust in background verification.
What connects all of those different industries is that they each deal with systems people depend on deeply, often without realizing it. There’s usually a very hard technical problem underneath, but also a big human consequence if you get it wrong. I’ve always found that intersection compelling and motivating.
The other consistent theme has been scale and transformation. I join organizations at moments where the technology, the business, or the operating model needs to evolve quickly. I’m energized by ambiguity and environments where you have to build structure while the ground is still moving underneath you.
At Mapbox, you moved through three distinct roles over nearly five years before leading the data platform team. What did that period of growth teach you about building systems that can evolve alongside a business, and what would you do differently if you were starting that journey again today?
I moved through a few very different roles there, and each one changed how I thought about scale. Early on, the focus was mostly on solving immediate technical problems and delivering quickly. Later on, especially leading the data platform teams, I started thinking much more about durability, coordination, and how to create systems that other teams could reliably build on top of.
One thing I learned is that scaling technology is really about scaling decision-making. As companies grow, more teams are contributing to the same ecosystem, and small inconsistencies start compounding quickly. You can’t rely on tribal knowledge anymore. You need clearer patterns, better interfaces between teams, and stronger operational habits.
That period also changed how I think about leadership. I realized that platform and infrastructure teams are at their best when they act as force multipliers. Their job isn’t just to build good systems, it’s to make the entire engineering organization more effective.
If I were starting that journey again today, I’d probably push earlier for simplicity and focus. I think as engineers we naturally want to build flexible, sophisticated systems, especially in fast-growing companies. But a lot of long-term scalability actually comes from being opinionated about what not to build yet. Some of the best decisions are the constraints you introduce early before complexity sets in.
You've spoken about the importance of discipline over motivation when it comes to leadership. Can you share a moment, in your career or as a leader, where that principle was truly put to the test, and what came out of it?
Discipline matters most when leading through periods of significant pressure, especially when the organization is looking to leadership for certainty that doesn’t fully exist yet.
In those situations, motivation comes and goes very quickly. People are tired, the problems are ambiguous, and there usually isn’t a clean path forward. What matters is showing up consistently, making decisions with incomplete information, communicating clearly, following through on commitments, and staying calm enough that the team can keep functioning effectively.
I’ve learned that during those times leadership is less about having moments of inspiration and more about creating stability during uncertainty. There were definitely periods in my career where I felt exhausted privately, but the responsibility of leadership is that your team still needs clarity, consistency, and direction from you.
What came out of those experiences was a much stronger appreciation for discipline and emotional steadiness. Teams pay attention to how leaders respond when things are hard. In my experience, discipline is what allows you to show up consistently over time.
Your current role puts you at the center of one of technology's most active battlegrounds: fraud detection in the age of AI-generated content. Your team is essentially building systems that have to stay ahead of tools their own engineers might be using to build faster. How do you think about designing for that tension between enabling AI-assisted development and protecting against AI-powered deception?
It’s a really interesting tension because the same technologies making my teams dramatically more productive are also lowering the barrier to sophisticated fraud. AI is compressing the gap between legitimate and malicious capability very quickly.
On the engineering side, we absolutely want to enable AI-assisted development. Teams can move faster, explore ideas more quickly, and reduce a lot of repetitive work. But it also changes the threat landscape because attackers now have access to many of the same accelerants. Synthetic identities, deepfakes, automated social engineering, document generation, all of it becomes cheaper and more scalable.
So the way I think about it is that trust can no longer rely on static signals. Systems need to become much more adaptive, probabilistic, and behavior-driven. You’re designing for continuous verification rather than one-time validation.
It also changes how you think about organizational design. Fraud prevention can’t sit in a silo anymore. Security, data, infrastructure, product, and machine learning teams all need tight feedback loops because the environment evolves too quickly for isolated ownership models.
The other important piece is speed. Historically, trust and safety systems were often reactive. In an AI-driven environment, reactive is too late. You need instrumentation, monitoring, and iteration cycles that let you detect emerging patterns quickly and adapt before they scale.
At a broader level, there will be a shift in how we think about trust online. We’ll have to design systems that evaluate authenticity continuously while still preserving a good user experience. That balance is going to become one of the next defining technical challenges.
