1. Could you walk us through the application of AI in the real estate arena?
I joined Redfin six years ago and founded our analytics engineering team. Redfin’s objective is to redefine real estate in the consumer’s favor. We’re using data and machine learning to deepen the relationship between our real estate agents and our customers. The very first thing we were looking at was how we could predict what homes our customers would be interested in, so we built an algorithm called, the “Redfin Matchmaker” and started making recommendations to our customers. We found that our algorithm was much better at predicting what homes the customer would like. It turned out that when Redfin recommends a home, customers are four times as likely to click on that house as they are on a home that fits the criteria of their own saved search. Also, pricing for a house is an emotional topic; someone lives in that home and has a connection to that home. We aim for our algorithm to be more than just a standalone number, to show users what homes were used to create that number, which is very important, as most providers give you only the figure. That is how we have tried to make it a part of the conversation and not as something that puts distance between the agent and the customer.
2. What sort of infrastructure do you feel should be used to power AI?
When we first wrote our algorithm years ago, we had tested it and were ready to turn it on full blast. So it was late at night, which was when cloud computing costs were cheaper and so we started to roll out that algorithm to run all of our data and a couple of minutes later, we got a phone call from AWS saying, “hey,do you mind turning off that algorithm for a few minutes, you are taking up too much of our computing resources right now!” This is funny now because AWS has added so much compute power since then, we would have a hard time gobbling it all up.
when Redfin recommends a home, customers are four times as likely to click on that house as they are on a home that fits the criteria of their own saved search
It showed what it was like, living on the edge for companies like us, that are pushing the limits of computing power, in order to get meaningful insights out of data, and so for us, using the cloud was critical.
3. Considering all the roles you have played how do you feel they have contributed to your current position as CTO?
I love data engineering and big data, but another thing I really enjoy is this idea of the efficiency of how you build systems that run efficiently which engineers can then build on top of. We’ve invested in helping engineers quickly set up their development environments so that there is nothing in their way when they sit down at their keyboard and start typing; we want our engineers to be as productive as possible. So as I have moved into the CTO role, I’ve worked to make sure our architecture helps our software developers be more effective and identify what is getting in their way.
4. Your role as being part of the women’s program in your company, could you shed some insights?
When I joined Redfin, I was the only woman on our Seattle software development team. Now, 30 percent of our developers and our technology team are women. Over these six years, we have become one of the few companies that have shown a meaningful change in the numbers. We’ve had to do a lot of different things to get to where we are now; I definitely think that bringing me on as a leader in engineering helped in the early days. Having a female leader can attract a lot more female engineers, but it is certainly not where we stopped. We also had to look at our hiring practices and our promotion practices, to look at how to build an inclusive culture where we make sure that all voices are heard, and it continues to be something we work on to this day.
5. What advice would you like to give an aspiring CTO?
As an industry, we believe in continuous improvement, where if we invest in people, if we give them the right training and the right opportunities, they can learn and become better software engineers, and I think the tech industry has believed in continuous improvement for some and not all. I think it is easier to teach people how to code rather than to teach them how to be a decent human being. It is important to look at everyone as an unfinished project, to look at yourself and the people you manage that way.