We told you why you should hire data scientists, now we’ll tell you how.
Whether you’re building a startup or trying to innovate at a well-established company, you’re likely always on the prowl for technical talent. While it’s hard enough finding skilled programmers, finding great data scientists – the unicorn employees of the tech world – can prove to be an even tougher challenge. To be effective, you have to determine right time to hire a data scientist, create a large pipeline of potential hires, and build a reliable and consistent interview structure. Here’s how you can apply tried and true growth hacking techniques to come out successful.
Understand the Market
The biggest reason it’s so hard to find data scientists is simple supply and demand. A candidate with with expertise across mathematics, code, and business is difficult to find, and with Glassdoor reporting more than 2500 data scientist jobs in San Francisco alone, demand is sky high. Anyone with a brain and a background in data science is being recruited non-stop, even if they aren’t looking for a new job. Knowing what you’re up against is the first step toward finding the right hire
Know When You’re Ready
People get into data science because they want to solve interesting problems. Outside of typical concerns like salary, vacation, and benefits, the thing that data scientists care about the most is what they’ll actually be working on.
Here’s what candidates want to hear about when they talk to you:
- Interesting and diverse data
- Tough problems that no one else has solved
- Mentorship with established data scientists
- The overall impact of their work
- Why this is a rare opportunity
Companies like Google and Facebook have a huge advantage here because the scale and power of data they’re able to offer data scientists. It’s unlikely that you’ll be able compete with them on that front, so make a strong case for why your data and the problems you’re solving are really interesting.
If you aren’t able to articulate exactly why you need a data scientist right now, you’re not ready to hire one.
Build an Engine
To cast a wide net and get as many applicants as possible, you need to figure out where data scientists hang out – both in person and online. Thankfully, this is actually the easiest part of the hiring process. Here’s a non-exhaustive list of some of the places you should check out during your search:
- SF Data Science
- SF Data Mining
- SF Machine Learning
- Bay Area Women in ML and Data Science
- Data Science for Sustainability
- Stitch Fix
Places to share your own content:
- Linkedin Groups
- O’Reilly Strata
- The Data Science Summit
Google is your friend. Search for meetups in your area, attend them, and build relationships with people in the local data science community to data scientists in their natural digital habitat. Just avoid being too pushy about hiring. Word gets around the data science community fast, so don’t screw up your reputation by acting like a sales rep.
If you’re looking to take your visibility in the community to the next level – and have the budget – you should consider sponsoring a meetup or networking event. This communicates to attendees that you’re serious about hiring and can put your money where your mouth is. As part of your sponsorship, negotiate with the organizers to do a short pitch about your company when the meetup starts, and ask if you can have your name and logo placed on any marketing materials.
The Interview Process
So you’ve determined to hire your first data scientist or grow your data science team. Most postings, in our experience, receive hundreds of applications for a single position. How do you find the best candidates?
First, offer a take-home challenge. It separates the most determined candidates from the pack and allows you to access their technical and communication skills. A great candidate will be able to take a small dataset and test several hypotheses, generate a basic model, and write up their results. Finally, have your team assess how the candidates use their knowledge of programming, statistics, and machine learning to solve the problem.
For those that make it past the first round, offer an in-person data challenge or case-based interview where the candidate actually works on something with your team. Companies that have great interviewing processes, like Airbnb and Uber, favor real-world challenges over whiteboarding interviews. And always have your team check for both technical fluency and soft skills. Even if they’re stellar at building data science products, they’ll need to be able to communicate effectively to make an impact at your organization.
There’s no need to reinvent the wheel when it comes to hiring. Data-driven companies such as Uber, Airbnb, Khan Academy, Mapbox, and Sailthru all structure their interview process similar to what’s outlined above. Copy the basics, then tweak it to best suit your team.
It’s hard but not impossible to find amazing data scientists. To give you a baseline ideas of how many people you’ll have to reach out to, here’s a rough estimate:
If there are 500 inbound applicants, 250 (50%) will submit a take-home test, 25 (10%) will pass, 20 (80%) will come to an in-person data challenge, 4 (20%) will pass the challenge, and 3 (75%) will accept an offer. That means in order to find a single great hire, you need over 150 applicants.
Start building a big pipeline and run with it. Best of luck!