When it comes to relative newcomers in the Data Science field, there aren’t many out there doing better than Alyssa Columbus. Although she just graduated from college earlier this year, she already has a full-time data scientist role at Pacific Life, a laundry list of conference and symposium speaking engagements, and has founded a local group for women who code in R. Oh, and she’s a NASA Datanaut.
In other words, if you want advice on how to break into the industry successfully, she’s a good person to ask. And while some of her advice will probably sound familiar, she also said some things you really might not expect.
Build the right foundation
Before you even think about applying for jobs, of course, you need to be sure you’ve got the skillset employers are looking for. And thankfully, there are tons of resources out there for people who want to learn about statistics, computer science, and data science.
“I advise you to [...] take some free online courses from platforms like Coursera, EdX, and MIT OpenCourseWare,” Alyssa said. Specifically, she recommended MIT and Harvard’s intro to computer science courses, MIT’s course on machine learning and statistics, and Stanford’s course on machine learning.
She also recommended Dataquest (we swear, we didn’t ask her to). “I’ve taken a few Dataquest courses and found them very helpful,” she said. “I took Command Line and Git and Version Control; I’d also recommend that.”
Once you have the basic foundation, the next step is to reinforce it with some on-the-job learning and here, Alyssa says, you can’t be afraid of internships. “Internships and research positions may eventually turn into full-time job offers if you work very hard and show initiative,” she said. “I worked very hard over time to exceed expectations at my internship at Pacific Life this summer, and that’s how I landed my full-time data scientist role. I also have previous internship experiences too; I had two or three internships in data analysis. That definitely helped me to gain more experience, skills, and expertise.”
Making yourself visible online
Having the right skills is necessary, but skills alone are never going to get you a job unless potential employers can see what you’re doing. “If you really want to stand out,” Alyssa says, “Have a very strong online presence, in the form of a website, portfolio, GitHub, blog, Kaggle profile, or all of these that showcase your interest, passion and proficiency in data science.”
“Your strong online presence will not only help you in applying for jobs,” Alyssa said, “but employers and others may also reach out to you with freelance projects, interviews, and speaking engagements.”
This article is an excellent example of that point. Dataquest reached out to interview Alyssa because we saw her posting cool things when she was rotating curator of the R Ladies Twitter account. Putting yourself out there on the web can be scary, but it can lead to all sorts of unexpected opportunities if you stick with it.
But what should you be putting out there? At a bare minimum, Alyssa says, you need a portfolio and some kind of website: “Get yourself a website if you haven’t already. It is essential for a data scientist in 2018 to have a website and a GitHub,” she said. “GitHub Pages are free to host your website. You can even write your website in R, and I’d recommend Blogdown for doing that.”
What makes a good data science portfolio
Everybody knows that having a good portfolio is important for finding data science jobs (here's our series on portfolio building, but what goes into your portfolio is really dependent on what you’re looking for. Alyssa recommended starting by researching the types of roles you’re looking for, and planning portfolio projects that demonstrate skills relevant to those roles.
“There’s kind of a balance that you have to strike in a portfolio,” she said. “You have to show expertise in a specific type of data, and also show that you can do work in a range of domains. When you’re starting to build your portfolio, think about four projects that you’ve already completed or could complete quickly that show both depth of expertise in that type of data and range of expertise.”
“Keep in mind that the most commonly-reported type of data is relational data for all industries, with the exceptions of academia and the military or the government,” she added.
You’ll definitely want to include projects that highlight your coding skills in one of the top languages requested in data science job postings (Python, R, and SQL, in that order). Make sure that code looks professional, Alyssa advises. “Have clean readable code, use version control, break your project into multiple files. Comment, comment some more.”
“Another type of project that I recommend is a visualization or storytelling,” she said. “Storytelling is a critical skill, along with communication.”
Your story could take the form of data visualizations or just a well-written write-up, but the end goal is to present your data in a format that’s both compelling and easy to understand. Even in projects that aren’t focused on storytelling, Alyssa recommends including a write-up. “Posting the code on Github is great, but it’s even better to have a write-up with it, whether it’s a readme or a blog post or something else,” she said.
She also recommended including a project that demonstrates your ability to deploy something, “whether it’s a RESTful API for a machine learning model you trained or a nice R Shiny or a Tableau dashboard. If you’re looking to get more into data science, I’d recommend R Shiny,” she said. “If you’re looking to get more into business intelligence or data analyst roles then I’d recommend starting with Tableau, especially if you’re not comfortable with coding yet.”
“Portfolios are also about what you don’t include,” Alyssa said. Your portfolio should be a distillation of your work that shows off your absolute best, so don’t throw every homework assignment you’ve made into it. It also needs to be a reflection of your skills, so group projects should only be included if you made a significant contribution to them.
She recommends not focusing too much on simple data cleaning or exploratory data analysis. Those elements can be included as part of other projects, but they probably shouldn’t be the focal point of any project you’re showcasing in your portfolio.
Finally, your portfolio should showcase unique projects that you’re genuinely excited about. In interviews, you’re going to be asked about these projects, and “your excitement will definitely make the interviewers more excited to hire you,” Alyssa said. “I’m speaking from experience, but also it’s common sense.”
How to actually find entry level jobs (hint: you’re doing it wrong)
Once you’ve got the skills and you’ve got the portfolio, it’s time to hit the tech job boards and start looking for cool jobs, right? Wrong.
“Don’t just look at job boards for jobs. When you’re job hunting it may be tempting to look for work on company websites or check specific job boards, but according to people who are employed in the data science industry, including me, these are among the least helpful ways to find work,” Alyssa said.
“Instead you should try to contact recruiters or build up your network to break into the field.”
“In order, I would look for work via recruiters, then friends/family/colleagues, then career fairs or recruiting events, then general job boards like LinkedIn, then company websites, and then tech job boards,” she said. In other words: job boards should be the last place you look.
Why? “A lot of those turn into black holes if your resume doesn’t stand out in any way,” Alyssa said. And because finding and submitting jobs on web-based job boards is quick and easy, your resume is going to be buried under a mountain of other applicants.
So what should a total newbie with no recruiter contacts or connections in the industry do? Of course, they should work on fixing that by going to meetups, conferences, and taking advantage of other networking opportunities. But Alyssa said that reaching out directly to recruiters you don’t know is OK, too. “I would reach out on LinkedIn. I would search ‘data scientist recruiter’ or something similar on LinkedIn and message them before asking to connect, or message them while asking to connect.”
The key is to remember that any relationship is a two-way street. “Don’t just message them ‘Do you have a job for me?’” Alyssa said. “Instead think of how you can make their job easier searching for candidates you know they need. Study their company and see where their greatest challenges are and explain how you can help solve them." That’s the quickest path to getting an interview, she said. "Conveying your interest in helping their business grow goes much farther than immediately messaging them asking for an interview." Alyssa said. "You get what you give, so show how you can make a contribution as best you can and hopefully an interview will follow."
Five key takeaways
If you’re looking for an entry-level job in data science, this article is full of gold, but here are five important bits of wisdom Alyssa shared to keep in mind before you file that next job application:
- Don’t look down on internships. Job experience is job experience, and they can lead directly to full-time staff positions.
- Put yourself out there, on the web and in real life. Your never know what jobs or other opportunities might come to you, but nothing will come to you if you’re not visible on the web and actively participating in the community.
- Tailor your portfolio to the job you want and the passions you have. You’ll have to talk about a lot in interviews, so make things you’ll be excited to talk about.
- Stay off job board sites. You’ll get better results by building relationships with recruiters, and leveraging connections in your network.
- Remember that relationships are a two-way street. Don’t just ask people for jobs, think about how you can help them out (even if it’s just buying their coffee).