Please tell us about your role and the team / technology at Oxygen.
I’m the Chief Data Scientist at Oxygen, so I oversee all data science efforts across risk, operations, marketing and product. My team’s vision is to empower the company with Data Science and Machine Learning so we can provide better business value with machine learning based features and cultivate a data-centric culture throughout the entire organization. As one of the first team hires, we made it a priority from day one to embed this into our culture, and it has given us a leg up, I think.
What was the idea behind starting a company like Oxygen?
I wish it was me who came up with the idea, but I can’t take credit. Our Founder and CEO, Hussein Ahmed, is a serial entrepreneur originally from Egypt. He sold a company there before coming to the U.S. to pursue a Ph.D. program in Computer Science at Virginia Tech. After a brief stint in engineering at Amazon, he founded and sold a collaborative software company in the Seattle area. While searching for his next idea, he enrolled into a part time MBA program at UC Berkeley Haas and started consulting in the interim. It was there the problem and idea for Oxygen came to him. Despite being quite successful and making a good income, he kept getting denied basic financial services and was finding his options limited to “alternative” lenders. “How could this be?” he thought. “I have great credit and a good income — why am I being shut out?”
He had office space at a WeWork with others like him — solopreneurs doing things ranging from design and development to copywriting and even music — so he started asking around if they were having the same issues. Turns out, they were. It seemed most traditional financial institutions were focusing on the higher end SMBs, and this new and burgeoning segment of the workforce – what some dub as the “Freelance” or “Creator Economy” – was being left out.
Look no further than the recent PPP struggles to highlight this mismatch. When he saw how big this new “21st century economy” was, with some estimates pointing to 90M independent businesses by 2028, he saw a massive opportunity, and thus, Oxygen was born.
What kind of challenges does your company solve for its customers?
Our customers face the same struggles most small businesses do and their concerns are largely financial. Things like simply being able to make and save enough to weather the storms of irregular income streams, saving for retirement, and accessing credit are all at the top of the list of concerns.
Considering this segment of the workforce overwhelmingly skews towards the artistic and creative fields, most are not immediately thinking about how to best run a business right off the bat. They get started doing what they are doing because they love it. They are not thinking “How can I best incorporate for tax and liability reasons” or “how can I maximize my cash flow?”
Over time they learn that to do all of these things and create a sustainable business they need to access multiple programs and applications to run the day to day operations of their business. Things like invoicing and payments, incorporation, expenses, accounting, insurance and core banking – these are all separate programs. We aim to make it easy to start and grow a business for this segment by offering a platform for all of these things in-house — giving them both the tools and education needed to thrive.
Fintech is one of the most proliferated data science marketplace. How does Oxygen use data science, analytics and AI?
As mentioned, Data science and AI have been part of our DNA from day one and are used across product, fraud and risk, marketing and operations.
From a product feature standpoint, we apply AI/ML to improve and enhance the user experience within our app, providing meaningful information to our customers. A few examples include enhanced transaction logic, where we take a messy transaction date and provide the customer with detailed information such as the merchants logo and location so they can better recall where they made purchases. Currently, we are building a model to detect recurring transactions of our customers. On average, consumers spend over $600 a year in subscription services, many on things they no longer use and remember they are paying for. By presenting this data, we can help our customers better manage their finances.
On the fraud and risk side, we are building several internal models to better detect fraudulent activities, protecting both our customers and our business. In Operations, we analyze multiple metrics to continuously improve the customer experience, which has led to strong customer service scores. On the marketing front, we analyze reams of data to better utilize different marketing channels — optimizing for lifetime value and acquisition costs — which has allowed us to scale significantly. We can already see our investment in this area paying off and we see this as a competitive advantage that will provide better network effects as time goes on.
What startups in fintech / other domains are you keenly following?
I’m personally interested in AI ethics and model interpretability. With the advances in AI and Machine Learning, the potential underlying biases are amplified and reinforced. As we all know, machine learning models are only as good as the quality of the data, so I’m always asking “How can we train a fair model with potentially biased data?” It is a critical yet challenging issue.
Another trend that fascinates me is model interpretability, as it is key to model utilization. Internally, operation, fraud, and audit teams will all be strongly empowered when we can provide details on how and why the model gives the result. Externally, interpretable models can also provide more transparency and assurance to customers.
I’m following a few startups that are aiming to enable and empower data scientists to be more productive and more powerful. Arthur.ai, for example, helps data science better monitor model performance over time. Dataiku is another one I follow closely. It enables data scientists to easily set up data pipelines from accessing raw data to pushing them to production, which is critical in collecting the vast array of data needed to train models.
Any advice to all fintech professionals or those trying to break into the space?
This is an area I am particularly passionate about, particularly for people in my domain of Data Science, and I regularly volunteer my time giving talks and tips to how best to break into the industry.
Generally speaking, domain knowledge is key. Understanding the statistics and machine learning algorithms is one thing, but how to apply the knowledge properly to solve real word problems is another thing entirely. Data scientists should obtain domain knowledge in the Fintech industry and understand the specific problems we are trying to solve before applying any Machine Learning models to the problems.
For early or even more seasoned professionals looking to break into the space, don’t be afraid to reach out! There are some basic rules of how to do so, but it is true that a warm introduction is better than a cold application as the majority of jobs aren’t even published. So get to know people in the industry and reach out for their own experience. I know this is hard for some but you will find it easier as you do it. People are generally open to talking about their experience and it’s a great way to make a professional connection. This is called an informational interview and it is important to not go into it with a transactional mindset — as in “help me find a job.”
Your goal is to listen, learn, and show remarkable things about yourself – make a good impression. It’s always nice to ask politely if you can follow up from time to time, and also let them know that you’d love for them to keep you in mind for any future, potential connections.
Then, follow up with them. Set up a google alert for a subject they spoke about and send them articles from time to time to continue the dialogue. It shows you listened and keeps you on their radar. It may not be tomorrow, but this strategy absolutely pays off. And make sure to pay it forward — if there is something of value you can provide, reach out. Networking is sometimes seen as a chore that isn’t at the top of a priority list. But I’ve found that the best networkers offer value. If you have that mindset you will stand out and be top of mind for when opportunities become available.
I am particularly grateful for my mentors who have helped me a lot in my own career progression, and I make it a priority to mentor others in return. Nobody gets to a position of success without the help of many along the way.
Thank you, Ivy! That was fun and we hope to see you back on globalfintechseries.com soon.
Ivy Lu, PhD, has joined the Oxygen team as Head of Data Science and Machine Learning. Her onboarding marks the launch of Oxygen’s banking platform, and according to Oxygen founder and CEO Hussein Ahmed, is indicative of how important data on customer experiences will be in informing Oxygen’s products and product development. “We can’t make our customers’ lives better if we don’t know their pain points – Ivy’s work will be crucial in that regard.”
Oxygen is a modern financial platform designed for the 21st-century economy – the digital-natives looking for a banking partner that understands how they live and work – providing a seamless user experience for both personal and business accounts in a way that makes them feel in their element. Available on iOS and Android, Oxygen users enjoy no monthly fees, no paperwork, early direct deposit, simple transfers, and cashback rewards on everyday purchases from approved merchants. Businesses can easily control their finances with integrated expense management and invoicing solutions that are elegant, simple and secure. Reject ordinary. Banking for the extraordinary, from Oxygen.