Fotis Konstantinidis joined CO-OP Financial Services in January 2018 as Senior Vice President overseeing fraud products. He sat down with Insight Vault to discuss data science, machine learning and fighting fraud in an always-connected, Internet of Things world.
Are there misconceptions about data scientists? What might people be surprised to learn about you?
When people hear you’re a data scientist, they assume you are a programmer or a big data engineer. In other words, they consider you more of a “doer” than a “thinker.” In actuality, most data scientists are focused on the thinking side of data initiatives. They are tasked with things like deciding which data is relevant, determining the best methods for collecting it and choosing the right algorithms and decisioning processes. They are the ones who figure out the most optimal way to uncover hidden patterns humans can’t detect on their own.
Ironically, though, my career in many ways perpetuates the programmer stereotype. For me, thinking through the strategy hasn’t always been enough. I typically feel compelled to see an idea through to execution. Early in my career, before big data analytics software had become somewhat democratized, I was writing code, testing algorithms and building models to see if what I had thought up would actually work.
You were a part of the innovation team that worked on connected car pilots at Visa. What were some of the key lessons learned or discoveries from that project?
When I first joined Visa, the card network was just beginning to open itself up to working alongside the Silicon Valley disruptors. The culture was changing, and my task was to begin to build relationships with some of these firms, as well as the many legacy companies within the connected car ecosystem. That included domestic and international car makers, oil and gas companies, QSR chains, Bluetooth, RFID and WiFi providers and so many more.
Although we had imagined a multitude of use cases for payments on board a connected car, we focused in on some of the most relevant for today’s consumer – things like automating gas, parking and drive-through payments.
In terms of my personal take-aways, three stand out:
- It’s always better to attack a problem from the perspective of the end-user’s needs rather than the latest and greatest technology.
- When you’re collaborating with many different organizations throughout an ecosystem, you have to understand how to speak different languages. What’s important to a car manufacturer isn’t necessarily important to a gas pump supplier.
- Just because a product or service might be cool, there has to be a supporting business model. At the end of the day, most of the parties in the ecosystem are going to want to see how the innovation either adds or saves revenue.
The project also solidified my passion for rapid prototyping. I’ve seen firsthand the value of this digitally transformed way of achieving innovation. It’s incredible the things you can accomplish when you don’t have to have all the questions answered before you begin to build.
What are the greatest challenges to fighting payment fraud in the always-connected IoT world?
One of the greatest is authenticating all the different devices already on the market or soon to be on the market. Machine learning is going to be a massive help to us in this regard, however.
Today’s fraud rules are of a human construct. Tomorrow’s will come from machines that have learned to think like humans – or rather, like super humans. And that’s what it will take to stop fraud before it starts.
Humans simply can’t keep up with the data input. Machines, on the other hand, can take in data from financial accounts, credit bureaus, social media, smartphones, GPS, connected thermostats, smart cars and more, all at once. Just like a human, the machine will make decisions and learn from them – except it will do it much more rapidly and in a way that is predictive and prescriptive, not rear-facing or reactive.
You have worked for both startups and legacy enterprises. In your experience, which has an easier path to meeting the demands of today’s digital consumer?
Startups are nimble. Legacy firms have resources. There are pros and cons for both.
The biggest advantage for a startup is that everyone working there is motivated to find a solution; they’re not making money until they do. The other thing they have going for them is that they’re typically using the latest tools, technologies and methods.
But, there are great pockets of research and disruption happening inside the Fortune 500 world. Our innovation group at Visa worked exactly as a startup would. It was a startup within a legacy organization.
What’s really interesting is the cross-pollination we’re seeing in the C-suite of both types of companies. Startup founders are entering the legacy world; CEOs of big companies are launching their own startups. CO-OP is a great example of a company that has a number of great leaders who have already executed on digital transformation. They understand how to modernize processes and systems to better meet the needs of modern end users.
What does digital transformation mean to you and how will you apply that definition to the strategy you develop within CO-OP’s ecosystem?
Digital transformation is about becoming more nimble by modernizing existing systems. Admittedly, that’s an oversimplification, but it beats some of the more esoteric definitions floating around on the Internet these days.
So why do we want to become more nimble? Because it’s easier to customize your products and services for clients. Whether you’re a credit union, an acquirer or a processor, the way to win with today’s customer is to offer personalized solutions that can be easily adapted to their evolving needs.
My job will be applying that transformation to the ways in which CO-OP helps credit unions detect and prevent payment fraud. Modernizing with machine learning will allow us to intervene and make new rules on the spot, with risk engines that are completely customized to each credit union’s unique fraud goals.
It’s obvious you have a passion for machine learning. What brought on your appreciation for this particular form of artificial intelligence?
I’ve worked with the technology in many different ways throughout my career. The most impactful, however, was my work applying machine learning to brain research. Our objective was to map different data points in the brain to help diagnose early stage Alzheimer’s disease. It was very fulfilling work that exposed me to the vast amount of good that can come from properly harnessing this powerful technology.
Too often, artificial intelligence is viewed as a means for replacing humans. My work in neuroscience proves it can be just as important to preserving us.