Advice for Aspiring Data Scientists
Pro-tips from a startup founder and ex-Airbnb Data Science Manager on helping your application stand out
A few years ago I wrote a piece called “Advice for New Data Scientists,” based on my time on Airbnb’s Data Science Team. The piece was one of KD-nugget’s most viewed blogs in April 2019 and it has nearly 40k views on Medium since then. It’s been syndicated in various places — and even plagiarized, if you can believe it! 😂 I’m wary of people who give too much advice, but I really like economies of scale and I have too many requests for this type of info that I’ve ignored (sorry!). So I’ve decided to try to synthesize my advice on breaking into your first data science role. This advice is based on nearly 5 years at Airbnb where I sat on dozens of hiring committees for the data science team and was a hiring manager for numerous roles, as well as my more recent experience as a startup founder and CEO at Iggy.
To set expectations correctly, this post assumes you have some training in data science or analytics. To my mind this means you can code in R or Python, swiftly transform data using common packages in either, write complex SQL, have a solid command of statistics, and sufficiently understand various machine learning techniques and algorithms. (For resources on getting up to speed on these topics, check out my former Airbnb colleague Kelly Peng’s excellent post). Most people think this basic training is enough to get a first job in data science. I am sorry, but that is largely wrong. That basic training is only the foundation. The tips I will share next are your springboard off of that foundation.
The Portfolio
If you take nothing else from this piece: build a portfolio, ideally in Github, that showcases your data science skills. It is my firm belief that this is the differentiator between any application and an application that actually gets attention. The goal of the portfolio is to bring your CV (resume) alive; it separates you from the hundreds, if not thousands, of other applicants for your first data science job. Your portfolio is where you showcase your work.
Some ideas for what to include in your portfolio: analyses, code gists, webapps, data documentation and blogs (+ README files!). You don’t need all of these by any means but if I had to choose two, I’d choose a webapp and accompanying blog post. A webapp is a great way to show your ability to link together different pieces of software and create something dynamic, hosted on the web. But why a blog? As I argued in my last post, communication is one of, if not the most important aspects of your job as a data scientist. Written communication is especially vital, and even more so if your job is remote. A well-written blog post (with linked code) allows the reader to get a sense of how you communicate, code, and think. If they get good signal from this, they will want to talk with you. This matters because getting your resume looked at is the hardest step in the job search process, so if you can increase your chances of conversion here, you’ll be in a great place.
You may now be wondering how to get inspiration for your portfolio. What about starting with a cool dataset you see referenced on Twitter or Kaggle? Are there any data quality issues like leakage, truncation, missing data? How do they impact an analysis? You could also get inspiration from newsletters from thought leaders (See the resources section, below). Choose some articles that interest you from current and past issues and read them, start following people on Twitter and you’ll soon have plenty of data science content coming your way. Some people will share analyses — see if you can replicate and update. Others may discuss a data science challenge that needs to be solved — how would you solve it? Still others will show they can do something in Python — can you do it in R? And finally, companies like mine will release new tools for data scientists. (Here’s another one that I dig). Try out their tools and share what you’ve built! You get the point. The purpose of all of this is to grow and learn your way into a job. It will take time.
The job search
Once your portfolio is ready to go, start looking for jobs! If you’re interested in startups, peruse TechCrunch to see what companies have been funded, and check out their job listings. Or do the same for the portfolios companies of various VC firms. I’d target companies that have raised at least a B round, though a C round may be a better bet. Why? Well, as a junior data scientist you’ll likely require a sufficient data infrastructure in place to be effective, and this is less likely to be the case for younger companies. If you’re more interested in jobs at larger companies, check out their job listings and learn as much as you can about the work they do. If they have a tech blog, read it. Look into what they open source.
Cold outreach
Those who know me well know that I am a fervent supporter of diversity in tech. There are many reasons I started Iggy, but one of them has to do with the chip on my shoulder from being a woman in a supremely male-dominated field. It is my firm belief that women and minorities should be building companies, driving decisions and engineering the best products imaginable — just as men do. Unfortunately, so many women (myself included!) wait to be told they can do these things. To share my own experience: I was intent on getting promoted as a data scientist before leaving data science to start my own company. I for some reason thought that I needed that stamp of approval to go and do something different (!!?). As soon as I realized (with some help) that that made no sense, I took the leap. Why am I sharing this here? Most men do not wait or ask for permission to do things; they just do them! As a founder hiring for many roles, I get cold emails nearly every week from men who want to join my company. I have received 0 cold emails from women. So, women reading this: learn how to send cold emails. This may mean getting slightly out of your comfort zone and applying for a job even when you don’t meet 100% of the listed requirements for the job. This is fine! Just lean into the portfolio you’ve built and make clear your complementary background or deep personal interest. This can offset a few of the minor job reqs. Men do this and don’t think twice and it helps them. This reality frustrates me, but as a founder I can tell you that hiring is really hard, so when people make my job easier by connecting the dots on their resume via a short cold email, I appreciate it.
I’ve included sample text for a cold email below. Please don’t copy it verbatim — use it as inspiration in writing something that reflects your voice.
Closing
I hope you’ve found this helpful. You will know you are doing a good job when recruiters reach out to you, and they are much more likely to when you give them a good reason to! If you have a question, ask it in the comment section. And if this has been helpful, please share your story below and let me know.
Lindsay Pettingill, CEO @ Iggy
Resources
Tools for your portfolio
- Github (to host all of your code): https://github.com/
- Heroku (to host webapps): https://www.heroku.com/
- Iggy (my company; for location enrichment): https://askiggy.com/
- Hex (for interactive notebooks/apps): https://hex.tech/
Newsletters
- Data Science Roundup http://roundup.fishtownanalytics.com/
- dbt community newsletter https://community.getdbt.com/newsletter
- Data Elixir: https://dataelixir.com/
- Projects to know https://projectstoknow.amplifypartners.com/ml-and-data
Sample cold email template
Hi, I just saw/read/heard {reference something substantive about company}. It was compelling/interesting to me because {connection to prior work, something in portfolio}. I’ve done some work in this area {link to it} and would love to chat about data science roles at {company}. Do you have time for a 15 minute chat in the next few weeks?
