Q&A with Mark Striebeck—Co-founder and Head of AI Engineering at Arta

Meet Mark

Co-founder Mark Striebeck brings decades of expertise in software engineering, technology systems design, and agile methodology to Arta Finance. Previously, he held senior engineering roles across world-changing applications, including Gmail, Google Research and Machine Intelligence, Google Payments, and YouTube. Currently, Mark leads all AI Engineering efforts at Arta.

Throughout your career, your teams built applications that feel simple and intuitive from the user’s view yet are extremely complex from an engineering perspective. Can you give us some insight into how you think about this balance? Maybe start with YouTube?

Even if you've never used YouTube, you intuitively know what to do with it when you land on the page. Yet, it is a very complex system behind the scenes. For example, YouTube has countless unrelated videos, from do-it-yourself tutorials to political speeches to the latest cosmetics reviews. The engineering team had the challenge of normalizing across these unconnected and diverse types of content. Ultimately, the team designed a user interface that hides all the complexity but still makes the functionality accessible and valuable. 

I was also responsible for community systems, including how users interact with videos, from reading comments to responding to them and sorting them. Comments, behind the scenes, require complicated engineering. Some users want to engage with the creator, media, and fellow viewers as much as possible. However, although commenting is a wildly popular feature, some users prefer avoiding interaction altogether. Those users simply want to watch videos. The challenge was to design a UI that worked for all of these users, no matter how they preferred to engage or not engage.

We segmented these user groups to understand their needs and interests in detail. Then we designed an interface that would work across all of those needs. So, yes, the YouTube backend is incredibly complex, but with an intuitive UI, it feels simple for the user/viewer. That is always the goal.

You were responsible for Gmail’s entire front end for mobile and web. What was the central challenge there? How did that compare to the challenges at YouTube?

At its core, Gmail is a productivity tool. Nobody goes to Gmail to have a fun, interactive experience. That's something you do on YouTube. People use Gmail because they need to get things done.

Whenever we had a big idea for Gmail, we had to ask ourselves the core question: how does this help someone go through their email better, faster, and with more engagement? If it didn’t accomplish that, it was not central to the user’s goal, and we did not prioritize it. That said, we had to find the right balance between functionality and engagement because if we had focused solely on function, using the app would have felt laborious and dull. On the other hand, if we focused on a highly engaging UI, we could have ended up with an application that doesn’t do much besides look cool. So it was crucial to strike the right balance, with a focus on the core priority, which was creating a productivity tool.

How do you determine the right balance?

It always comes down to understanding what the user wants and needs. Then, we must break down these complex constructs in the background and figure out how to expose them to the user so that they can accomplish their task without needing to understand the complexity required to pull it off on the backend.

For Gmail, did you use a similar segmentation approach to that at YouTube?

Yes, we did. Many Gmail users, first and foremost, want an efficient experience. They want to quickly read their emails, preferring to spend little time in the app. Others live in their email and want to see everything with detailed control over sorting and organization. We wanted to create a universal UI that works for everyone.  When Google started Gmail, we created four distinct user profiles. We ran every new feature through those four profiles to ensure that every change meant something to each user group and didn’t get in the way of the others. Understanding the user and running tests was absolutely essential.

Can you talk more about the importance of testing? You lead the effort at Google to change the company’s testing infrastructure. What prompted that effort? What was that experience like?

I was responsible for Google's test infrastructure for a couple of years. We completely changed how Google did software testing during that time by building the world's most extensive continuous integration system. 

For some background, it’s customary when developing software to write code and then run tests to check that code. Testing is crucial to ensure the code’s quality, performance and reliability before putting it into widespread use. Moreover, tests can reveal when new code accidentally breaks code somewhere else in the system.

Typically, companies use this process where there are one or two thousand lines of code. Google has billions of lines of code. The challenge at Google was that there was one repository for the entire company. Everything from ads, search, Gmail, YouTube, maps, and the infrastructure was in one single repository. So, when an engineer made a tiny change in the storage infrastructure, they could simultaneously break a feature in Gmail by accident. There was a risk with every code change, and tests were instrumental in revealing the issues as early as possible.

Before implementing the change, teams would test their code in isolation and sometimes learn that another team’s changes accidentally broke their application. At a tech company with many teams writing code simultaneously, you can imagine how complicated it was to contend with code fixes constantly. We needed a way to make testing easier and reduce the number of surprise accidents. We had to scale and parallelize the testing and implement the best of computer science in every way.

The result was a system that, within a few minutes, tells you if your new code breaks anything, anywhere at Google, across the entire repository. Whenever someone wanted to submit new code, they could first check to see if it would break other things. If it did, they could fix their code before implementing it. It was game-changing and defined a new level of best practices for testing software.

The overarching goal was that we wanted engineers to write more tests. Furthermore, we wanted testing to be an approachable and meaningful experience for the engineer - something engineers wanted to do. The new system was one-of-a-kind machinery that elevated the overall quality of work to the absolute highest possible standards.

This sounds particularly relevant to your current work at Arta. 

Yes, definitely. At Arta, our entire system is built and held to the absolute highest standard possible. Security, stability, and precision are paramount when it comes to finance. Particularly for Arta’s trading system, top-quality code and infrastructure is essential. We are working with people’s hard-earned money to help them grow and manage their wealth. It requires the highest standard of engineering at all times, in every way.

We are working with people’s hard-earned money to help them grow and manage their wealth. It requires the highest standard of engineering at all times, in every way.

How do you set up the team for success with these high standards?

I'm drawing a lot from my time at Google when I had to figure out how to move the entire company to build at a new level. At Google, we needed “change agents” to catalyze the transition to this new approach. Shifting culture takes a long time, so it was a massive effort. With Arta, we are setting the highest bar from the very start.

Exceptionally high code quality should be focused, not stressful. I’m dedicated to creating a company culture and work environment where it’s easy to take the approach that will yield the highest quality software, driving engineers to write high-quality code and tests. 

We are also fortunate to have extraordinarily talented engineers at Arta who are committed to exceptional quality. 

I’m dedicated to creating a company culture and work environment where it’s easy to take the approach that will yield the highest quality software, driving engineers to write high-quality code and tests. 

Can you tell us what you and the AI team are currently working on? 

In addition to the trading system I mentioned earlier, I’m responsible for Arta’s underlying infrastructure for all the data we collect. This includes market data, economic data, and everything we need to build our AI models.

We are building an all-encompassing data model that aims to combine and normalize across all asset classes and more. You might think of it as a systematic, digital version of everything a responsible, high-performance financial advisor typically does in their head with a far smaller data set. 

To create an overall data model, we must do everything 100% systematically and apply the most advanced computer science and information technology. We have an accomplished team, and managing this massive amount of data, calculations, and variables is something that plays into our strength as a company. 

From a technology perspective, what Arta is doing is complex, but this will be behind the scenes. Like at YouTube, Gmail, and Google, the user experience will feel smooth and intuitive. We’re drawing from vast experience to apply all we’ve learned from finance, computer science, user experience, and interface design to help expose complex financial concepts to the user in a way that they intuitively understand and draw the best conclusions to meet their goals.

We hope to make the complex process of wealth management feel easy and accessible for our members. It’s a step change in managing finances, and we’re thrilled to share this with the world.

It certainly sounds like complexity is a significant theme in your work and career. What do you do to relax and recharge?

Well, continuing on the theme of complexity, you might not be surprised that one of my hobbies is astrophotography. It combines my passions - mathematics, astronomy, and computers. Does that mean my leisure time is distinguished by complexity too? Maybe, but the outcome is beautiful pictures, so the stakes aren’t quite as high.

I balance all that complexity by relaxing with my family, snowboarding, or scuba diving - nothing complicated about that!

BECOME A
MEMBER