The divide between research scientists and data scientists isn’t clearly defined. Still, research scientists are generally responsible for building and developing models in a very niche field using complex machine learning, mathematics, and deep learning theory.
While data scientists and adjacent positions work with broad swaths of data and metrics, research scientists encapsulate a deep understanding within a specific area of expertise, working cross-functionally with different teams to develop prototypes and validate hypotheses.
The requirements for research scientists largely vary by company: some have more stringent standards, while others are more flexible on the number of years of experience, degree levels, etc. Even within companies, job qualifications can change between different teams, so double-check for positions you’re interested in.
Generally, you can be sure to expect the following:
For some companies and positions, there’s also an additional, rather unique, requirement:
To illustrate the sheer variety in research scientist roles across different companies, here are some examples of open positions in top-tier companies within tech and other industries. Please do note: several of the given examples of research scientist positions at these companies are hiring at the mid to senior level, so the requirements listed may not reflect the qualifications needed for research scientists as a whole.
Lyft has several open positions around the world for the general Research Scientist position. Within this role, you’d work on different teams with engineers and product analysts to analyze data and provide business insights.
This company has rather stringent requirements, looking at candidates with an M.S. and at least four years of industry experience and a whole slew of additional preferences.
In addition, positions with specific designations and divisions (such as “Research Scientist, Autonomous Driving”) work with small, highly specialized teams to develop certain products and services, often within industry-breaking roles. These jobs have higher requirements: a Ph.D., deep learning knowledge, publishing experience, etc.
Compared to Lyft, which has a variety of general Research Scientist positions in addition to specialized ones, Facebook organizes all of its research scientists into specific divisions. Here are some examples currently open worldwide:
Depending on the area, these positions often require a Ph.D. in a relevant field, at least a year’s worth of lab experience, and first author publications, along with extensive experience involving different concepts, skills, and techniques, depending on the position.
The Linkedin research scientist role is comparable to a combination of Lyft’s general research scientist responsibilities and Facebook’s hiring requirements.
The research scientist job at Linkedin consists heavily of data manipulation and organization, with variable allowances for specific teams. Candidates are generally expected to have an M.S. or PhD in relevant fields, with several years of industry experience depending on the specific role.
Other qualifications primarily depend on the team you’re applying for. For instance, the AI team requires in-depth machine learning knowledge for developing algorithms within the job search function.
As always, there are several key steps in the interview process for a research scientist position. Here’s a broad framework of what you can expect, though it definitely varies by company.
This interview will be conducted either with a recruiter or hiring manager. General topics to be covered will include your past experience, your resume, and certain projects you’ve worked on that relate to the position. This is a very general ‘getting to know you’ conversation between you and the company, so just relax and put your best foot forward.
This is (most likely) another phone interview where you’ll be expected to demonstrate technical knowledge specifically pertaining to the research scientist position. Common subjects will most likely be tied around machine learning concepts, analyzing case studies, and basic statistical concepts. The difficulty and breadth of the questions in this interview depend on the position’s focus. Note: some interviewers may combine the initial phone screen and the technical interview or simply skip the former altogether, so make sure you’re always prepared!
The on-site interview is the last part of the interview process for research scientists. At this point, you have passed the basic screening and adequately demonstrated your technical knowledge, so this interview is more of an assessment of your fit within the company and, if relevant, within the specific team.
You can expect lots of behavioral-type questions, as well as potentially more technical questions relating to the team’s current projects. Here, the technical interview will be less focused on the range of your knowledge and more on your approach and justifications for specific parts of the problem.
Culture and fit questions are sure to come up, so it’s a good idea to research company values and really learn more about the goals of the team before going into the interview.
Finally, the structure of the on-site interview largely depends on the company itself. Larger, more-established companies, like Facebook, Amazon, Google, etc., likely have multiple interviewers and rounds for the on-site part, meaning you may be interviewing most of the day with different members of the team, other research scientists, and so on.
What are the types of interview questions for research scientists?
How does the X concept work? / What are the assumptions of X? / How would you apply X?
You get the point: this type of question basically asks for a textbook recall of different machine learning concepts and their applications or conditions.
Don’t overcomplicate it! Interviewers here are just checking that you know and understand basic concepts. For research scientists specifically, you can definitely count on basic machine learning concepts to come up at least once. Other common topics can range from different statistical applications to programming questions.
Case study interview questions may be asked during the technical screening of research scientists’ roles but almost definitely within the on-site interview.
Here, the focus isn’t so much on the specific methodology or content (although you don’t want to be spewing out total nonsense), but more on your approach– the choices you make towards different features and their trade-offs. If you can justify your approach and actively talk through the problem with the interviewer, you should be good to go.
This type of question requires more individual research into the specific team and company you’re applying for. The structure of this question will follow along “How would X change as Y changes?”, where X and Y may constitute different variables in team projects.
Basically, this question just really comes down to the commitment in your preparation. The interviewer is looking to evaluate whether you know the goals of the company, the focus of your team, and the overall objective of your position.
Coding questions come up pretty frequently for research scientist positions. Almost all research scientists at tech companies have to develop and simulate their ideas and theories in practice before handing them off to an engineer to actually scale.
Most research scientists’ coding interview questions revolve around testing mathematical and algorithmic concepts. You could be asked to simulate different kinds of random variables (binomial, bayes) or simulations (monte carlo).