How To Interview and Hire For A Data Scientist
Hiring and Interviewing A Data Scientist
Before we get started looking at how to hire a the right data scientist. Let’s look at a great example of what not to do when hiring someone!
We hope that brought a quick smile and maybe a laugh to your morning or afternoon. Now on to hiring that data scientist or starting that data science team.
As with any hiring process. Finding a data scientist that is both a good fit for your company culture and meets the technical requirements requirements of the job specifications is not easy.
One of the biggest issues facing most companies today is that data science is still a relativity new field. It may seem like everyone is doing it based off all the hype. However, general companies not focused as highly on technology have just recently started investing in data scientists.
Many are just starting initiatives, and data science teams. Our team just recently conducted a workshop at a multi-billion dollar company who was finally considering starting a data science team.
Part of their questions were, how do they hire a data scientist, can they just hire their own scientists, do they need to hire PhDs?
We have also have been asked questions like:
- How do we know what skills you are looking for?
- What projects should data scientist be doing for the best ROI?
- How do we know if we have found a good data scientist?
- Who should manage our data science team?
There are plenty of questions your team could ask. We don’t recommend you Google what a data scientist is. Not because it will be wrong. It may not be accurate for your data science team or your business. Just like programmers, not all data scientists have the same skills and backgrounds.
Technology is changing so rapidly, some businesses that are new are able to use new technology Stacks, and data storage systems. However, some companies are so large that they are still running much older systems. Some systems aren’t running on all the fancy storage systems that Google and Amazon operate on.
That doesn’t meant they can’t approach data science problems! They just need to hire data scientists who either understand their systems, or who can quickly learn them.
Here are a couple of tips and questions we know will help your leadership team create a job description and interview for your new data science positions. If you have specific questions, don’t hesitate to contact us.
Questions To Ask Your Team To Help Develop A Job Spec
What Projects Is Your Data Science Team Working On
Like we have said a few times. Data science is a broad field (Some might argue that some of our examples are more specifically machine learning, which is not incorrect. we are just discussing the broad field of data science).
Thus, the types of projects data scientists do are also broad. It applies to chat bots, deep learning, product recommendation, strategy development, new value streams, and so on! So what project is your company looking for a data science to do?
If the data science project requires terabytes of data that stream live, then you will need a data scientists that can manage Hadoop or some other multi-distributed system.
There is no way SQL can truly handle that data well. Trust us, we have seen companies wait days for queries to return because they were managing such large data warehouses.
If on the other hand, your company is managing data warehouses that have data-pulls occur infrequently, then hiring a data scientists with a SQL background will suffice.
From a coding perspective. We believe that data scientist should be fluent in either R or Python and at least aware of how to use the other. R is more of what we consider a research language. Even though SQL Server has recently integrated R into SQL Server 2016, its not at the level of system integration that Python has.
Python works so well everywhere, that when it comes to integrating code and models into larger systems. Python works great(of course you could also learn C).
What Skills Are Needed On The Data Science Team? –
Data science is a field with several disciplines intertwined. Machine learning, statistics, and predictive modeling are just a few of them. Some data science jobs require software engineering experience, others statistical modeling and still others require knowledge Hadoop and hardware. Not all data scientist can do everything. This is ok! As long as you know what specific skills you need from them.
Some data scientists are purely engineers. If you want them to build an algorithm or integrate some module into an old system that still runs on Cobol and you don’t have an engineer on your team. Then your job specification and interview questions should be geared towards a software developer with a hint of machine learning.
On the other hand, if your team is seeking a member that can develop new value streams, and has a knack for talking to executives. Then state that clearly on your Indeed.com ad.
This all depends on the data science project you would like to start.
If this is a new data science team. What projects do you want them working on?
There are plenty of projects most data scientists are capable of doing. Many of these projects could also be done by an analyst. This can save anywhere from 20-50k depending what company you are at.
The average data scientist makes upwards of 120k in Seattle. Where as many analyst make somewhere in the range of 70-100k(Analyst is a much broader term, we are assuming this is a SQL, and report heavy analyst).
Data scientist should be working on projects that require heavy amounts of statistics feature engineering, and scientific like research. That means proving a conclusion several times over, testing hypotheses, some software engineering and having a disciplined approach to problem solving.
Make Sure The Candidate Believe In Your Companies Product or Service
When we think of data scientists, we think of passionate individuals who are hungry to find value and solve problems. It is key trait of data scientists. Not just coming in and performing basic analysis, but actually searching for ways to make both their team and their company better. If a data scientist isn’t excited about a companies product, or questions the companies vision. They might not be the best fit (of course that could be said about any employee). Data scientist in particular should be interested in the problems they get to solve. It’s what we love most. It’s the problem we get to solve, the impact we have on more than just our department. There is something satisfying there.
Move Quickly When You Find The Right Candidate
If you have found a possible data scientist who your team enjoyed and you believed had the right technical and soft skills you are looking for. Then we recommend you get to them as soon as possible.
Good data scientists move quickly off the market. Other managers at other companies probably want them as well. So make sure your company provides certainty early. Sometimes that is all a new hire is looking for. A certain job(of course, some will use this to their advantage and ask for a higher salary). There is always some trade off.
Some Coding Questions Are Great, But Don’t Rely On It –
We have seen great data scientists not get jobs because they might not spend all their time practicing interview questions. Coding interview questions are another skill that has to be learned. This is not the same as actually programming. So there can be a lot of false positives.
We do recommend some standard questions to make sure they could code their way out of a paper bag. However, don’t expect them to develop your next chat bot right there on the spot! With that point, don’t ask for the equivalent of free labor. It is tempting. You could get a cohort of candidates to essentially write a few modules for your data science project. Just coming from a data scientist…it feels tacky and a little unfair.
Data scientists are not just programmers. They are problem solvers, they are data driven and tinkers. This isn’t always best judged by code. We recommend you ask them to go into detail about their last projects with another technical person. There is only so much you can BS.
Look for business experience, not just a PhD-
PhDs are great! However, an employee with very little business experience doesn’t always provide that much value. They will need to be partnered with someone who can help them use their immense knowledge towards business problems.
We were reading another article that described this same issue. They used terms like “Lack of entrepreneurial spirit”, which is another way to put it. This we think depends on the role you are hiring a data scientists or machine learning engineer for.
If the position is more focused on engineering and system integration, then that person doesn’t need to require as much of a business sense. However, if you are hoping to have a data scientist that can search out new value streams, push strategy, and drive new product ideas with data. Then you want a data scientists that also has strong sense of what drives a business.
There are few out there that have both strong business sense and strong technical skills. As mentioned prior, most teams are typically built up of multiple types of data scientists. The key is to know which ones you need and when.
Ask Them Specifically How They Approached Some Of Their Past Problems –
No matter the background. A data scientist is a problem solver. Ask them how they solved a specific problem on their resume. Drill into it for a little bit. Ask them about their data science process, the problems they ran into, was there missing data, how did they deal with any external politics. Data science is not just crunching numbers, it is about dealing with limitations, bad data, internal politics, etc.
If you need to, have a technical person present to ensure that the answers are not simply BS.
Look For Bright Data Analysts and BI Engineers Internally Too! –
There are a lot of brilliant analysts and BI Engineers who are probably well suited to start being trained for junior data scientists roles(or citizen data scientist). Companies are always so eager to hire externally, vs. training up.
However, guess what! The best companies out there are just setting up their own data science universities! Why? Because they don’t have enough talent being pumped out of schools! Check it out, Amazon and Airbnb both are doing it!
Why? Because if you have strong experience with programming, or data engineering, it is just a few steps away form becoming a Jr. Data Scientist. Yes, these analysts have a lot of experience they need to build. However, they can maintain code, build basic modules and clean data.
All of this is grunt work that takes about 50-80% of most data science teams time. Not something you want your Sr. Data Scientists doing! Try looking interally!
That all depends on the projects your looking to get done, and the technologies your company as.
Data science is considered hot right now and companies are still trying to figure out how to hire the correct data scientists(even Google is still trying to figure out this problem). We recently held a workshop at a multi-billion dollar company how data science was being used in their industry and how they could start to implement their own data science team.
Many companies are just starting as companies like Amazon begin to threaten every industry. Companies are realizing that there is something to this data science and machine learning. Amazon’s edge isn’t magic. It is a combination of pushing its employees to be the best, and forcing decisions to be data driven first. Part of that, is hiring and setting up a data science team!
Good luck finding your team of data scientist!
If you liked this read please feel free to check out some of our other data science posts.
Hire Data Scientist