Secure your business from login to chargeback
Stop fraud, break down data silos, and lower friction with Sift.
- Achieve up to 285% ROI
- Increase user acceptance rates up to 99%
- Drop time spent on manual review up to 80%
Technology moves fast, and Fred Sadaghiani has always kept pace with it. Fred thrives in environments where he can solve hard, fun, meaningful problems. This drive has led him to startups like TeachStreet, to behemoths like Amazon -- and now to Sift, where Fred is CTO.
At Sift, Fred spearheads the development of creative and dynamic technology. How did he devise his innovative approach to building systems? What lessons did he take away from massive companies at the cusp of change, and from scrappy little startups trying to build something new? What is it about this company's machine learning tech that makes Sift so successful? Fred answers these questions and much more.
Evan: Welcome to “Trust and Safety in Numbers,” presented by Sift Science. I’m your host, Evan Ramzipoor. About six years ago, when Sift Science was still a young, scrappy little start-up with wild aspirations, Fred Sadaghiani joined the company as a CTO.
Fred: Always been, kind of, in the technology and internet space.
Evan: This episode is the first in a two-part series. In today’s podcast, we’ll hear about how Fred developed his innovative approach to building systems. In the second part, Fred will talk about what’s on the horizon for machine learning technology, and how it’s changing the very nature of the internet. But first, let’s warm up with a quick fraud fact.
Did you know that payment fraud costs the travel industry an estimated $858 million a year? To learn more, check out our fraud in the travel industry infographic on the Sift Science blog. Now, onto the interview. So Fred, tell me a bit about yourself.
Fred: I started my career in 2001 working at Amazon. It was kind of the very early days of the company when obviously it was a much smaller place than it is right now. But more importantly, it was just very focused on books, and music, and video, and DVD. And it’s kind of where I, like, caught my teeth on systems, and engineering and really kind of like learned to build products and see a company grow. And after that, I had a couple of stints at various start-ups. I worked at a company called Zillow, it’s an online real estate search engine.
Evan: That was where Fred met Jason Tan, Sift Science’s CEO. So they’ve known each other for about a decade.
Fred: You know, when we started at that company, it was very unclear what it was. And we, again, got to see, or I got to see explosive growth. A company that was just kind of a hair-brained, kind of exciting idea but really connected with its audience, and I think transformed the real estate space in a very meaningful way.
Evan: After that, Fred spent some time at a start-up called TeachStreet, a marketplace that connected teachers with students and vice versa. Usually, so that students could learn cool things that you can’t always learn in a classroom like painting or yoga.
Fred: Another one of my start-ups was a payments company and we built payments APIs for game developers. And this was kind of back in the hay day of flash-based games where a lot of time and a lot of users, and more importantly, a lot of money was being spent in buying virtual goods and virtual currencies. And kind of our value was that we had a native payments API.
Evan: It was there, in a world where payments and gaming collided that Fred started to learn a lot about fraud.
Fred: You’d be surprised how much fraud there was in the world of virtual goods and games. And we ultimately ended up at Google. And I was at Google for a couple of years, again working in the payment space. After that, Jason and I talked, and you know, I decided to, kind of, jump ship, go from the big company back to where my heart was, where I really got to connect with the problem, I got to connect with the users and the customers of what we were building. And I’ve been at Sift, yeah, ever since.
Evan: What is it that motivates you? What gets you excited to do what you do?
Fred: There’s a couple of things that motivate me. I think, you know, for me, I have a framework in my thinking, which is I wanna work on a hard problem, a valuable problem, and one that’s really fun. For me, you have to have all three of those things, and when you have those three things, it’s kind of like this magical scenario where you can kind of really be doing something that’s rewarding and fulfilling and impactful to the world. And so, you know, working at Sift, I feel really, really fortunate to, one, be working on a really hard problem, finding fraud is not easy. We have this problem where there’s, you know, 1 or 10th of a percent of the bad actors online who are ruining everything for the 99 or 99.9% of us. So there’s this class imbalance, so from a technical perspective, it’s a really, really hard problem. And then in terms of the value, you know, we’ve heard time and again how we’ve impacted not only the businesses that we work with and made their lives so much better as far as being able to, kind of, like forget about fraud and abuse as a problem in their view, so that they can continue to innovate and provide value for their end users. But also, for the users who are indirectly benefiting from not having to worry about fraud, not having their accounts taken over, not having their stolen credit card bills racked all the way up and dealing with those problems. So it’s really, really rewarding to know that we’re working on something that’s making lives better for people. And then the fun is, you know, getting to work with all the people that I do, who are just as passionate, and just as committed, and just as excited to make this impact that we’re working against. For me, that trilogy of things is what makes for a really motivating and rewarding work environment, and investment in my time.
Evan: How have you distilled what you’ve learned from your diverse experience at large companies and at start-ups into a strategy for approaching engineering projects?
Fred: Yeah, so, in the case of engineering, I think I learned a lot of really interesting lessons, you know, going all the way back to the days of Amazon. We really didn’t know what our customers wanted, what end users wanted. We tried this kind of build whatever we can, throw it against the wall, see-what-sticks approach. And in our mindset, in our approach, was this notion of iteration, getting customer feedback, and acceptance testing and then iterating, iterating, iterating. So very early on, I kind of learned this lesson that it’s kind of naive and premature to assume that you understand what the true pain or what the true need of the customer and end user is. And you really have to iterate. And you have to set yourself up from a systems and engineering perspective to allow yourself to iterate, test, and evaluate what you’re doing.
Evan: Fred says that means when you’re building something new, it helps to start with a holistic, high-level view of what you’re trying to accomplish, rather than trying to build smaller, complex components all at once.
Fred: And then building the lite version or the MVP version of that so that you can test your hypotheses as quickly as possible. And so you can really then once you start to see traction and see kind of positive response, deepen the investment across any part of that infrastructure or system. Whether it’s the backend database, or the middleware, or your API design, or whatever your systems are. Those things are all going to evolve and you have to think about what you build in having that forward evolution. And so, if you start with that mindset of, “Hey, whatever we’re building today is not going to be the solution for tomorrow,” you can really set yourself up for this iteration and kind of validation of your approach. And to be specific about that, you know, a company like Amazon, when I started there we had, I don’t know, probably a hundred computers that served all of amazon.com and the whole marketplace.
Evan: Think about it. There are millions of servers powering Amazon today. They only had about a hundred at the time. That’s nothing.
Fred: Well, those systems that we built years and years ago have all been replaced. They’ve all been rebuilt. And the nature of engineering, I think for, you know, those of us who have been lucky around, lucky to have seen this kind of dramatic growth and this dramatic scaling is that the solution you apply today won’t get you very far. You’re gonna have to reimagine it, and understand it at a 10x scale, and then 20x scale, 50, and then probably 100x scale in a very short period. So the half-life of any infrastructure is pretty, pretty small.
Evan: Let’s switch gears and talk about machine learning specifically. People familiar with the podcast probably know that Sift Science uses machine learning solutions to fight fraud. But why machine learning? What is it about that particular technology that makes Sift’s product so successful?
Fred: I think you have to kind take a step back and look at what Sift Science is trying to accomplish. Our mission is really to protect the internet from all different types of evil and bad actors. And the best approach to doing that is by looking at all the data we can possibly see from the different actions that end users are taking. The idea being that it’s not a credit card or a particular log-in. It’s not like a discrete event that is causing the fraud. It’s kind of the whole, broad behaviors that the users are taking that capture the true signal.
Evan: Fred says that if you start from the premise that it’s this background wealth of information that informs the diagnosis of so-called “bad actors,” then you need a system that’s designed to sift through all that data. Yes, the pun was intended, of course. And the only system that can possibly work at that scale is machine learning.
Fred: It’s the only approach that works at the scale where you’re looking at all the data. And so our philosophy has always been that the raw data encodes the truth behind what is good and what is bad. And if you believe that, then the only reasonable approach is by applying machine learning to the task of surfacing the good and the bad.
Evan: That was Fred Sadaghiani, CTO at Sift Science. On the next episode of this two-part series, Fred talks about how machine learning is changing the internet itself. Until then, stay vigilant, fraud fighters.
Stop fraud, break down data silos, and lower friction with Sift.