When it comes to building out a tech stack, engineering teams consider a wide range of factors, from cost to scalability. At fintech startup Dave, a strong tech backbone is built upon the need for speed and simplicity for a very specific reason: Their customers can’t wait.
According to Executive Vice President of Engineering Chien-Liang Chou, the organization’s tech stack includes Node.js, React, Python, Fivetran and Snowflake, all of which make it possible for his team to stay nimble and drive innovation. With this technological foundation in place, Chou and his teammates created a new underwriting engine powered by machine learning that had a major influence on the company’s ability to more effectively and rapidly serve customers.
“Our company was founded to help the financially vulnerable — those who may not have a great credit score or are living paycheck to paycheck — achieve greater financial health,” Chou said. “This means our underwriting engine is extremely important, as it helps us collect a different kind of data set about our customers that allows them to qualify for advances or learn how to better manage their budgets, for example.”
For future engineering hires, the breadth of Dave’s technological capabilities also offers countless opportunities for growth and influence. According to Chou, engineers who join the company can expect to leverage the latest technology in an effort to continually meet the needs of the business and pursue greater transformation in financial services.
“Because we have numerous new products and initiatives on our roadmap, there are always opportunities for engineers to learn something new and tackle different challenges,” Chou said.
Built In LA caught up with Chou to get an inside look at the company’s tech stack and learn how his team leverages it to drive innovation.
What tools are included in your tech stack, and how do these technologies enable your team to bring about transformation?
Grounded entirely in the Google Cloud platform, Dave’s engineering team abides by a specific philosophy: Keep things simple and move quickly. We built our own machine learning underwriting engine and the data platform to serve customers. We use Python for machine learning, data-building tool Fivetran for our data pipeline and Snowflake for data warehousing. Our back-end is developed with Typescript, Node.js and MySQL, and our mobile app is developed using React.js and React Native.
Leveraging a fully automated continuous integration and development environment and production observability system, our goal is to create a highly productive environment that allows our engineers to quickly develop, test, deploy, observe and iterate. These tools allow us to work quickly while supporting our customers.
How does the company’s newest underwriting engine serve customers more effectively, and why was this an exciting project to work on?
When Dave was launched in 2017, the company used a rule-based core underwriting engine. Using our data, we came up with a host of rules that would determine whether a member received a cash advance. As we continued to iterate, we started comparing the differences between a rule-based engine and a machine learning-based engine. We hired data scientists to fuel the next generation of our underwriting engine and realized that investing in machine learning would help us issue advances more effectively. For the next few years, we focused on investing in our data infrastructure and underwriting engine and continued to iterate on this model. This has been exciting, as we’ve seen significant progress so far.
What sets Dave apart is that we don’t ask for a customer’s credit score in order for them to receive a cash advance. By enabling customers to connect their bank accounts, our underwriting engine can analyze their transactions and understand their income. In doing so, it allows us to help them budget and provide an advance if needed. We’ve been able to move much faster and help significantly more customers while leveraging this model.
Why is it important for your team to leverage simple, scalable and iterative architecture?
We’re in a very competitive market. Dave was one of the first organizations to support a consumer base often written off by big banks. Now, we’re seeing older and larger banks eager to enter this market, while many other neobanks, or fintech firms, continue to emerge as well. While we’ve been able to grow quickly, we still have to keep our eye on the ball to continue leading the market. At the same time, we need to keep our system and architecture simple and clear in order to scale quickly. Many companies’ systems become way too complicated, which causes productivity to drop. A legacy system is hard to replace, and we don’t want to fall into that. Moving quickly enables us to move forward.
CREATE ECONOMIC OPPORTUNITY
How does your team decide when to add new technologies, and why is it important to remove old tech in the process?
Every company I’ve worked with has had a legacy system and each one wants to build the next generation of systems to fulfill future business needs. When building or adding new technologies, it’s important to think about how it will deprecate the old tech. This process can be likened to maintaining one’s home. If you keep buying new things without cleaning out the old, it can become disastrous, and you’ll reach a point where you have too much.
I always tell my engineers that if there’s a new and better way to do something, do it. Yet I also want to see how they plan to deprecate the tech they’re replacing. Adding new architecture is great, but if we continue adding without addressing legacy systems, then we’ll fail. Many companies don’t do that because it becomes expensive to deprecate. However, our philosophy is that it’s always better to iterate and trim down now before it becomes too late.
Can you give me some examples of how your team has leveraged new technology and the impact it had on your processes?
Our core system was built and iterated upon within a single code base. As we grew, it became monolithic and difficult for an expanding group of developers to work on together. We decided to break it down into microservices piece by piece, which allowed teams to move faster and build new products in much cleaner ways.
Using GraphQL also improved the team’s processes. Prior to its adoption, our developers proposed its usage to our architecture committee, giving them an idea of how it would make data retrieval more elegant, secure and efficient. The committee loved the proposal, and the team is currently building the pilot.
We are in a space where everyone can spread their wings and work in different areas.”
How does Dave’s engineering culture provide opportunities to use the latest tech and tackle challenges on a companywide scale?
For engineers at Dave, understanding business priorities and staying up to date with tech trends go hand in hand. We are in a space where everyone can spread their wings and work in different areas. We’re always looking for problems to solve while researching the best technology to address each issue. In doing so, we push our engineers to do their own research on how to conquer roadblocks.
.jpg)
.png)
