Faces of Uncountable
Meet Data Scientist Yijun Guo
Yijun talks the transition from academia to industry
We’re happy to share a series of interviews with members of the Uncountable team. The purpose of these is to shed light on the people behind our platform, including their professional backgrounds, what led them to join our team, and what they work on behind the scenes.
Today’s interview is with Yijun Guo, Data Scientist at Uncountable.
Employee profile: Yijun Guo — Data Scientist
UNC: How would you describe your job at Uncountable?
Yijun: I think there are two or three main parts of my work: One is to make sure the data on the platform is formatted in a correct way or in a clean format. And also, I get on calls, either sort of like a training call with customers, guiding them through how to use our platform, or more modeling oriented, which is just how to use the model. It’s more like training, training related, like, basically how to use the platform.
UNC: What do you like about working at Uncountable?
Yijun: Because I’m from a pure academia background, it’s honestly the type of data I can finally transition to a data scientist type of role. I think it’s more like the type of work, the type of data. But since I’ve been here for two years, the longer I stay, I feel like it’s not only about the type of work, but also the team, the people who you work with. I’m always feel like support it, when I have a problem. there is always someone kind of like can help me if there’s a bad day, I have like four customers requests that I can’t really finish, I can always find someone engineering team or someone else to help me out there.
UNC: What’s a particularly memorable part of your work at Uncountable?
Yijun: We were working with a customer who already was somewhat skeptical about our modeling strategy. And also the data they provide us on the platform isn’t really clean. So we went back and forth three or four times to make sure the data is clean and structured in the correct format. And towards the end, it’s already been, three or four months, to kind of make sure the data is clean. I feel like the customer is getting impatient. On that last round, I was like, do I really want to insist on like, getting all those things clean and structured? Are there other ways to do it, but honestly, there isn’t other ways to kind of like work around it. And I do feel like getting data clean and structured is something that I should insist on. So I did. After we structured everything, we saw some interesting trends. And we only ran like one round of experiments, less than 10 experiments, and the customer got what they were struggling with for a long time. So yeah, so I felt like it’s kind of like a valuable lesson. Because getting the data clean, and structure itself is very hard. And especially sometimes you deal with like this, sort of like impatient. And a lot of the pressure about like, is this worth it? Should I really spend so much time on this. But I feel like after this experience, I am like now I’m more confident to talk to customers, when their data is not clean. I always try to be like, there is no point of doing the modeling, when your data is not clean, or it’s not structured.
UNC: How has the way you think about your work changed since joining Uncountable?
Yijun: What I was a chemist in the lab, I definitely trusted my intuition more. But after coming here, working as a data scientist for two years, it’s not like I don’t trust my intuition, but I kind of want to see the real data, what the real what is the real data tells me? Is there something there? If there is not, it’s probably that my intuition is wrong. Or, maybe I just didn’t have enough data to support my hypothesis. So yeah, I feel like looking at the data, my mindset changed from like, you know, this drive to this because there is like a chemical reaction going on, more towards like, do I see a trend in the data? Is that trend reliable? Yeah, things like this.
This transcript has been condensed and lightly edited for clarity.
