In an effort to prepare students for the job market, we hosted an alumni panel discussion in our NYC classroom last month, during which three recent graduates: Lyle Payne Morgan Smith, Data Analyst at BuzzFeed, Erin Dooley, Research Analyst at NYC Department of Education, and Gina Soileau, Teaching Assistant at Metis, spoke candidly about their job searches, interview experiences, and current positions.
See below for a transcription of the discussion, which offers perspective and insight into the data science job search process. It was moderated by Jennifer Raimone, Metis Career Advisor.
Jennifer: Tonight, we really want to talk about how Metis has prepared you all for the job search, for landing a job, and for working within a data science department or on a data science team.
Why don't we begin with this question: how did Metis help prepare you for the role you're in now?
Lyle: I'm a Data Analyst at BuzzFeed. Before coming to Metis, I was basically a business analyst for a consulting firm focused on media.
Metis gave me the analytical tool set and the technical tool set I needed. And really, though I don't use that much machine learning in my job right now, understanding it allows me to have conversations with people who are using it, and helps me understand when it could be applicable.
Erin: I came to Metis from working in Broadway theater. I was doing ticket pricing, so I was using a lot of data, but everything was in Excel. I feel like I had some great ideas but didn't know how to implement them. I thought, "It would be cool to do this project, but I don't know how." So I came to Metis looking for exposure to the tools that are out there, and also, just in general, exposure to what the data science landscape looks like.
In my new role at the New York City Department of Education, I'm a Research Analyst and I feel like any idea I have, I know how to start implementing it.
Gina: I also have a Business Analyst background; that's what I did at a hedge fund for years. Before that, I came from a computer science background, so I was on the technology side. I was really interested in concepts and business flow. Then I moved to the Business Analyst side with the technical background, which is a bit unique.
As for what Metis has done for me...when you start looking on job boards, everything you've done at Metis is there. The skills all map perfectly. I was just telling Jason [Metis Co-Founder] that applying for jobs before Metis and after...it's day and night. What I'm qualified to do and what I can speak to now are just completely different.
Jennifer: Why don't we talk about final projects, Metis Career Day (during which hiring companies attend students' final presentations), the job search, etc. Let's first start with your experience at Career Day. What went well? And what didn't go so well, if anything?
Lyle: My project looked at DonorsChoose. I had a very strong belief that there was a pattern, meaning there were specific things that helped a project to get funded or not. I was not right about that. There were all sorts of external factors that I wasn't able to account for.
I built an app where you could put in a project idea, and it would provide the user with a percentage chance of whether or not it would get funded. I gave my presentation, and it was fine. I ended up talking not about how great my model was, but about the different impacts of the variables. Some things negatively impacted the chances of a user getting funded and some things positively impacted it.
I was stressed going into Career Day, thinking, "Oh no, they're going to ask me how my model performed. I'm going to have to say it's not great." But no one asked me that question. If they had, I would have told the truth, but I think a lot of data science projects take a long time and then end up not being what you expected. And that's okay, because you can learn from that, too.
A lot of people at Career Day just want to talk to you about your experience and want to get to know you a little bit and understand why you did this thing. They want to know: what was the passion that drove you to look at this project, and what did you learn from it?
Erin: Getting up there and talking, for me, was the hardest part. I think in preparation, I kept telling myself, "If I'm talking to an employer and they're asking me questions and I don't know the answers, or I've never heard of what they're talking about, then the job is probably not the right fit for me." I think it's more important to connect with them and have an interesting conversation about the project and not about each statistical, tiny little detail and technique.
Gina: For my project, I was trying to predict appreciation values for New York City real estate, in all boroughs and all neighborhoods. The thesis was, if you were house hunting and picked a house in a neighborhood that would appreciate the most, you'd get the most value. So if that was your metric for success, if you were an investor for example, then this was an application for you.
It wasn't exactly what I wanted, but you reach a certain point when you have what you have and you focus on your presentation because people who are watching won't care that your model got this much more or this much better. They're going to care how it looks, that you have a good front end, that it does something that's interesting to them. So the presentation should be just as key as how your model performs.
Jen: Excellent. Now, let's talk about the job search. What was your perception of the job search versus what it was really like?
Erin: Well, I really shouldn't be talking right now because my perception of the job search was that it was going to take 6 months and that it was going to be awful. I thought I was going to be crying every day because I needed a job and I was running out of money. But I got hired from a conversation that I had at Career Day. I had an interview 3 days after, then I got the job and now I work there, so I really shouldn't be talking.
Gina: I took about a month off. That was my preparation for the job search. I've had great experiences. I've only applied to about 12-15 places so far and I've approached it selectively. I've put together really nice cover letters that don't make me seem like a robot.
I spoke to some people who applied everywhere and they went at it really hard right away. I've seen people who've applied to 40 or 50 places and that has got to be hard because writing cover letters is challenging. I think it's important to pick what corpus of data you want to work with and then maybe that can help narrow your search.
Lyle: I take job hunting very seriously. I hate writing cover letters, so in order to avoid that, I asked people to coffee from various companies. Then you don't have to write a cover letter because coffee gets you past that first hump. I had a lot of coffee. Like maybe 30 coffees with anyone and everyone...people who came in to talk at Metis, people I saw on LinkedIn, and I used my college network a lot.
I'm not going to say what I did was a good or bad idea, but this is what I did: the first week out of Metis, I had 10 interviews. I've looked for jobs in the past and I didn't get the hit rate that I did when I was coming out of Metis. It was a much higher hit rate, which meant that I was booked. I was going to three interviews a day and a lot of them were technical. I wouldn't recommend going to three technical interviews in a day. That's a lot. Even if you're really on top of your game, that's a lot of game to have.
On top of all this, I kept a list of the 10 things I needed out of a job. When you get to the round when you're like, "Oh my God, they're giving me a job. This is awesome," it can be hard to ask yourself: "Is this checking off all the boxes on my list?"
You should be thinking about how much money you really need to be happy, about if you want to work on a team, or are you fine working alone? Do you want to work for a consumer-facing company? Do you want to work in a huge corporation or a start-up? Thinking about these things before you get into the search process makes it easier to ensure you find the right fit.
I got a job offer a week after Career Day based on someone who saw my presentation. I knew it wasn't the right fit, but I kept trying to convince myself that it was. The reason I didn't take the job was because it didn't check off all my boxes. I know it's really competitive out there, but you should try to make sure you're doing something that you like and that you're excited about.
Jen: I would like to explore those 30 coffee dates, because I think it's important to realize that applying solely online is not necessarily going to get you in the door.
Lyle, after doing it and reflecting back, what is the best approach? Were you talking to people who are data scientists, senior data scientists, heads of departments? Does it matter?
Lyle: I didn't really discriminate. I think one of the things you'll learn as you start applying for jobs is that titles are really hard to interpret. I don't know how it works at other firms, but at BuzzFeed, I am technically a Data Analyst on the data science team and I am just as capable of putting your resume through as anyone else.
Gina: Another thing I've done is gone through LinkedIn to see if there's a job that I'm interested in. I've then made contacts with people at the company and then maybe even sent them an Inmail [through LinkedIn] to say something like, "I saw your posting and I thought I would reach out," etc.
Jen: Great. Ok, I want to open it up to questions from the class now.
Student: Erin, what sort of work do you do at the Department of Education?
Erin: I just started about a month ago. It's really cool and really applicable to what I learned at Metis. We're working on taking a student in the education system and finding the 50 students most similar to them who are one year ahead of them. Then we use that cohort of 50 to set targets for the specific student.
It's really cool because from day 1, they had this Python script, this R script that they'd been using. They had some variables they had put into it and they turned it over to me to improve it.
Student: This one is also for Erin. I was curious, are you working as part of a team, or are you a lone data scientist, or somewhere in between?
Erin: I'm part of a team. I don't think city government is going to call it data science, but I work for the Office of School Performance and it's very data-driven. Some of the people on the team come from a SaaS background, some of them have a little bit of Python, a little bit of RSO. We're all kind of in the same boat in terms of learning together and figuring stuff out.
Student: Say you have a problem and you're trying to approach it and you come to a brick wall. Maybe you've entered all your data into something, or you have some kind of model and you're hoping for an outcome and it ends up just being noise. How would you overcome that?
Erin: I would say maybe go back and ask yourself, "Why am I even trying to answer this specific question?" Maybe the question is not something that can be answered with data and maybe there's something that you could use as a proxy to answer the real top-level question that you're looking into.
Gina: Definitely take a look at what features you're using and try to make those better.
Student: You mentioned you made a list of things that are necessary for you to take a job. How would you guys recommend we go about figuring out what those important factors are for us?
Lyle: That's a good question. I've had a lot of jobs, considering my age. I've had six, and so I have a lot of experience knowing what worked and didn't work for me. As for what that means from a data science perspective, it helped to talk with people in the profession.
Also, I knew that I didn't want to be by myself and I wanted to be part of a team that could help me, especially because I didn't have a strong technical background. I knew that I needed a job where I was going to be using Python every day or else I would lose that skill.
Jen: This is also something that we'll talk about in the one-on-one meetings with each of you post-graduation, so you won't have to figure it out alone. We'll work together with you.
Student: When you guys were interviewing for data analyst or data science-focused jobs rather engineering ones, did you find that there were a lot of technical questions focused on computer science theory and data structures? And if so, did you do a lot of additional preparation outside of the exposure that you got while at Metis?
Gina: I haven't had an interview like that. Most of the ones I've had are really SQL-heavy. If people are throwing random computer science stuff at you, you have to remember that you're choosing a company, too. It goes both ways. If they're just trying to get you with computer science trivia, that would be off-putting to me.
Lyle: I did get a lot of SQL questions. I would know SQL, just because people do use that. I got weird questions about making a recursive function and more computer science-y questions and I just tried my best.
For people interviewing at BuzzFeed, they ask serious data science questions and they're not necessarily interested in if you know the right answer. They're interested more in how you think about it and how you can talk through it. So many people are continuing to learn on the job all the time, so you don't go in knowing everything, but it's good to have the confidence to say, "Actually, I haven't used that algorithm before, but this is how I would think to use it." I think people respect that.
Jason Moss [Metis Co-Founder]: To what extent did you talk about your prior work experience separate from Metis during interviews?
Gina: For me, it's been a ton, because I have a computer science background. A lot of places are interested in hiring data engineers or someone coming from computers, like an engineer who learned data science. So they're focused on people who can build things and have business acumen. They were really excited about the business analyst experience. But I also applied to Johns Hopkins, and they wanted someone with a PhD and an academic background. So it will depend, because not everyone will look as much on the business side.
Erin: I spoke about my prior career in theater a lot when I interviewed. Of course, you don't come into Metis knowing nothing. You come in with a background and then you can use what you learn here on top of that to boost yourself up and get double points.
Jen: Thank you so much. This was a great conversation.