How hard is it to find a data science job? I love talking to current data scientists and machine learning engineers who have successfully navigated the data science job market after coming from non-traditional backgrounds.
Today, I wrote up my interview with Jerry, a former data science boot camp and master’s graduate who went through a ton of interviews before landing his current job as a machine learning engineer.
Hi, I’m Jerry, a machine learning engineer at U.S. Bank. I first started out working as a data analyst for a few years before getting my M.S. in Data Science at the University of New Haven partnered with Galvanize.
I knew I wanted to go more into the machine learning route, but it was hard getting interviews after the program. Eventually, I landed an internship on the analytics team of the Denver Broncos before finally accepting an offer with U.S. Bank.
My current role is more software engineering heavy but includes a good amount of data knowledge. A lot of building machine learning systems and infrastructures means knowing how to apply engineering efficiency to serving models in production.
I ended up interviewing for five months, from April 2019 to August. Some weeks, I would just have one interview, and other weeks, around 7–8 interviews. It was a pretty exhausting process, and I think I ended up talking to over 50 companies.
I ended up doing so many interviews because I felt like data science was such a hard subject to study, and I didn’t know what to focus on. Some companies asked questions about software engineering class structure and object-oriented programming, while others focused more on statistics. But I felt like I generally passed around 95% of my recruiting and hiring manager screens but only around 30% of my technical screens and take-home challenges.
The main problem was that each data science interview was so different. I had one interview for an Analytics Manager role where I interviewed with six different people before going to the onsite! I literally had to talk to the VP of product, hiring manager, general manager, engineer, data scientist, etc…
“Given there’s so many different ways to do things in SQL, even if you do it one way, interviewers will sometimes ask for a more efficient way of writing it.”
The take-home challenges and technical screens were initially really difficult. Data scientist is a very broad term that could encapsulate pure analytics, machine learning, applying deep learning, or being more business-facing.
I feel like since each position is different, just understanding what to expect was the hard part in the beginning. Interview Query’s question bank definitely helped me see the commonalities in each question for different companies and cleared up how to efficiently spend my time studying.
SQL questions in interviews were also my weak point for a while. I had always done stuff in Pandas and used SQL to pull data out to manipulate in Pandas. So, the Interview Query solutions were really helpful in terms of understanding how to think and structure SQL problems.
Given there are so many different ways to do things in SQL, even if you do it one way, interviewers will sometimes ask for a more efficient way of writing it.
Ultimately, interviewing is such a grind, and practicing was the best way to get ahead and get prepared for each interview. I got better as I did more interviews and practiced more problems.
Interviewing is a real grind. Especially if you’re new and don’t have a lot of experience, once you have experience, it was different because even though I felt like my skill set was pretty similar, because I had an extra year of experience as an intern, I would advance to the next round much more easily.
Being able to really talk in-depth about projects that I worked on also really helped cement my past experience. Hiring managers want to know that you can impact the business and make it better. But as a new person coming into data science, you don’t have many ways to showcase this skill set, which is why the technical interviews are so stringent. Lots of companies are more scared of false positives rather than false negatives.