For better or worse, there is a perception that people working in data have a fairly homogeneous career background—a long academic track, heavy on statistics, that has culminated in plenty of research and a PhD. Though this is certainly one route to a career in data, it is far from the ONLY one. As you’ll hear from Mike Tamir, Head of Data Science at Uber ATG, there are opportunities for people with non-traditional backgrounds to forge their own path into an amazing data career. Indeed, this broader pool of talent is really needed to keep up with the surging demand for data scientists at companies around the world.
Mike sits on our Data Advisory Board, providing expert input and advice for our School of Data Science programs. He also regularly recruits into data roles at one of the most disruptive companies on the planet. This is his advice on how to get hired in a data role without a “traditional” background.
What do you look for in a new hire?
I like to build data science teams that are pretty well-rounded. This means looking for candidates with technical skills in machine learning, software development, statistical modeling, and probability theory. For my teams, I’ve found that hiring a combination of data scientists, data engineers, and software developers makes for the best results.
Beyond technical data skills, what other qualities are important for you when you’re recruiting an entry- or mid-level role?
To operate as a high-performing data scientist, you often need to think like both a scientist and a product manager. So at Uber, we make sure every candidate has competency in experimental design and product sense.
What are your thoughts on recruits that come from a nontraditional data science background? For example, those that have built their data skills through independent study, or through practical project experience.
I’ve had good experiences hiring alternative path data scientists. Typically, these are students who had a technical academic background and a strong mathematical literacy, before taking the initiative to take alternative data science training, or they are people who had previous experience as a software developer.
So how does such a candidate impress you during the recruitment process?
Candidates who are able to walk through their actual experience of working on an end-to-end project often impress the most. This is especially true when they can identify where they made mistakes, and they can talk about what they learned from those mistakes. The ability to do this, while also checking the boxes on technical competencies, is a good signal for us.
What advice would you give job seekers who are starting the search for a data role?
Create a portfolio of projects that you can share. Especially if you’re searching for your first role, it is important to substantiate what you are able to do. The best way to do that is with real work. Also, you learn a lot from every project that will prepare you for the technical checks of your skills that a good interview process will have.
What is your top tip for a new analyst or data scientist in their first weeks on the job?
My number one tip is to always ask questions. A healthy data science and machine learning development culture should never discourage team members from trying to understand as much as possible!
Thanks, Mike, for sharing your expert perspective. If you’re ready to launch or advance your own career in data, take a look at the programs in Udacity’s School of Data Science. Whatever your background and starting point, you’ll find Nanodegree programs where you can master the latest data skills and techniques, and where you’ll build your very own project portfolio.