At Interview Query, we love chatting with Success Stories: members who have gone on to land data science jobs at top companies. We recently chatted with two data scientists who work at cryptocurrency exchanges in the U.S. and Mexico.
These interviews answered a lot of questions about data science in cryptocurrency, as well as how to land a cryptocurrency data science job. Here are some of the questions that were answered:
To learn more, we interviewed Sarah, who works as a data scientist at a U.S.-based crypto exchange, and Manuel, a data scientist at Bitso, a Mexico-based crypto exchange.
Here’s a quick look at the day-to-day work of data scientists at crypto exchanges:
SARAH: Within data science, you have got a few functions, one of which would be marketing. Within the financial space, you have questions around risk and fraud. But, more specifically to the cryptocurrency industry, there are questions specific to blockchain technology itself.
So it’s a lot of analytics type of work day-to-day. I build data pipelines, perform exploration and data gathering, and sometimes do some data cleaning. But eventually, it will be more analytics-type SQL queries, building dashboards, machine learning, and modeling.
MANUEL: My day-to-day has to do with modeling. I build predictive models that provide helpful insights to the company. So I get to do a lot of different things, like building random forests and linear regressions, to statistical analysis, like hidden Markov models or time-series analysis. But there’s a level of autonomy. I have a lot of projects on my to-do list.
SARAH: For the team that I work on, we do deal with questions that are specific to cryptocurrency. Although you don’t necessarily need cryptocurrency knowledge, it would be more challenging without it. You would have the simultaneous learning curve of the tools and processes in addition to learning crypto terminology and technology to even understand the problem you’re trying to solve.
But I don’t think it’s necessary to understand crypto deeply to work for a cryptocurrency company. There are data science roles in cryptocurrency companies that are similar to other industries. Those roles don’t require deep knowledge of crypto. For example, there’s always going to be marketing analytics functions and the need for optimizations.
So I think it’s an advantage for some specific roles, but you could work in the industry without it.
MANUEL: In the crypto industry, there’s a lot of data, and there’s a lot of room for data scientists to help a crypto exchange understand all that data and pull out insights. So no, you don’t need in-depth knowledge of crypto, as there are many business-focused roles that you’d find at any company with large volumes of streaming data. But it would help.
Interested in landing a data science job at a cryptocurrency business? Here are some tips and advice specific to the interview process:
SARAH: In terms of specifics, there was a take-home assignment, which isn’t unusual in the tech industry, and then the onsite after that. I enjoy speaking at conferences and have given talks about my past work over the prior two years, so I was already set up to talk clearly about my favorite past projects, which is important for any job interview that isn’t your first job.
I think a good tip for Data Science interviews is to practice talking about your work with people who aren’t familiar with it. It doesn’t need to be at a conference. It can be with a coworker, a friend who might understand your work, or even your mom, who has no idea what you do. Explaining my work to people with different levels of experience has really made interviews a lot better for me.
MANUEL: I had three interviews, starting with an initial screen that primarily covered statistics. I’ve since asked the interviewer: why did you focus on the basics? And he said it was a good test to see if a candidate really understands what’s going on under the hood.
The second interview was with engineering, and it covered basics like cloud computing. The third was with the CTO of Bitso. It was more of a general scope of why I wanted to work there and how I could help Bitso move forward. It wasn’t really a crypto-focused interview, although I did get a question like, “What’s bitcoin?”
SARAH: Interview prep is not something I’d ever done because I’d never worked for a tech company. I really wanted this job though, and I knew that interviews in tech companies were more standardized than what I was used to, so I wanted to learn that process.
For me, I had spent four years in a previous role, and I could talk all day long about the specific work that I had done. But I didn’t use K-means algorithms or build linear regressions in my day-to-day, which are very foundational data science and machine learning concepts.
The recruiter had told me before the onsite that they would ask about modeling, probability, and statistics questions. This is when I found Interview Query. I knew it was important to know what topics get asked frequently and what language they want to hear at a technology company. For example, A/B testing is a term that’s used widely in tech, but statisticians would just call that a specific type of experimental design. Coming from a non-tech background, it was important for me to understand the language but also what the minimum expectations were in given subjects. Interview Query was really perfect for defining what I needed to brush up on.
MANUEL: I had been working at smaller startups, and Bitso was starting to grow. So I was a little nervous about interviewing with a larger firm. But I found Interview Query, which helped. I wanted to practice a lot of different skills, especially Python. Up to that point, I was working in R mostly. And Interview Query helped me work on Python and learn the interview language.
But I also brushed up on statistics, took a look at my college texts and notes, and focused on the basic concepts that come up in interviews.
SARAH: I think, for data science in general, it is always good to know your statistical fundamentals. By that, I mean a basic understanding of probability and knowing basic statistical models. It’s obvious when someone tries to show off their project that uses very advanced machine learning algorithms, but doesn’t understand what a simple logistic regression is actually doing.
Also, you should be able to explain to an interviewer why you’re doing something even if it doesn’t seem hugely impactful to understand very deeply. For example, it’s easy to de-value data cleaning. You think, “Oh, it’s janitorial work. It’s boring.” But if you understand how to do that effectively, why you’re doing it, why it’s important, or how it’s going to mess up your algorithm, that goes a long way in interviews and in your career. It helps you to become a more trusted data scientist.
As far as the specific role I’m in, I think the best thing you can do is to start playing with cryptocurrency.
MANUEL: When we interview for data science positions at Bitso, especially at the junior level, we want to understand if they have the basics down. You might be able to build a really cool model, but do you understand what’s working behind the scenes?
For instance, if you are doing a regression analysis, if you only tell me that you predict that the value is X but can’t tell me what the most significant variable is, you’re missing the type of knowledge that we’re looking for. So if you want to do well, you really have to know what your models are outputting and why. You can’t just know the final output; you should know all the steps that led to that output.