How data science university recruiting works at Microsoft
By Archana Ramesh, Tom Mereckis, and Amy Siebenthaler
Data is integral to how we run our business at Microsoft, and data scientists within Microsoft work on a wide range of applications across many product areas (such as Windows, Xbox, and Office, among others), cloud systems, operations, and many more. We hire data science candidates with a desire to build products and make tangible business impact with data by applying statistics, Machine Learning (ML), experimentation, and many other related techniques.
Data scientists at Microsoft spend their time building and prototyping data science solutions, measuring and tuning performance, collaborating with other disciplines, and influencing for impact — all while scaling these solutions to a customer base in the order of millions. We focus on a wide variety of problems. We apply data science to gain insights into our customers, competitors, and our own business. We help build innovative data systems and tools at massive scale that thousands of our colleagues across Microsoft depend upon. We use ML and other data science techniques to improve features for our customers and partners. We look for ways to think broadly and uncover opportunities that span teams and divisions. We also partner closely with various feature teams to help those teams become independently skilled at data science and engineering — teaching others while simultaneously doing work that drives business impact.
What data scientists do at Microsoft
You might be wondering what a typical day in the life of a Microsoft data scientist looks like. It’s tough to describe a single day since our projects and the tasks within each project are so varied. But we can describe a typical data science project to capture a view of our work: As a data scientist at Microsoft, the projects you own tend to be roughly four to six weeks in duration. Here is also an example of a data science project at Microsoft.
Most projects start with understanding the problem area and framing the data science questions. This requires a solid technical background and strong communication and collaboration skills, as well as the ability to quickly learn and become an expert in a specific part of the business (e.g., product telemetry). This is followed by data preparation, including writing code to extract data, working with database systems and cloud storage, and then shaping the data for the analysis. After this, you start with exploring the data and structuring it in a way that’s suitable for analysis (which could include feature engineering). The first stage is typically both collaborative and iterative, as you work very closely with domain experts.
After this, you start figuring out the specific modeling choices to use. This includes leveraging both the data you’ve prepared and domain experts to answer several questions, such as:
- Is this an unsupervised problem or supervised?
- What are the approaches to try?
- What are the other considerations to factor in — does interpretability matter? Or is scalability of concern? In the latter case, is sophisticated modeling needed or are simple statistics adequate?
At the end of this stage, you usually reach a specific approach to take for the problem and start prototyping the solution. Within Microsoft we typically have flexibility to use a wide variety of tools for modeling (including R, Python, and so on). An important aspect in prototyping is evaluation — including being able to convince yourself and the key stakeholders that your solution works well for the problem. You then iterate and refine your solution, and then present your solution and/or insights to stakeholders. Next, stakeholders use your project to either drive a business decision or turn your project into a system that operates in a production setting or prompts other actions. During the course of the entire project, it’s typical to lean on a strong community of peer data scientists via peer reviews.
Data scientists at Microsoft usually enjoy tremendous flexibility in the areas that they work on. So, when the project is completed, you either move on to a new area or continue to follow on in the same area.
If you’re an aspiring intern, you might want to know the difference between what we’ve described above and a typical intern project. At Microsoft, we craft intern projects to closely resemble projects that full-time data scientists work on, but there are a few nuances. Because an internship is a finite time period (typically 12 weeks), sometimes an intern project might not be fully completed by an intern, and so might be handed off for completion to a full-time data scientist. Another difference is that an intern typically needs and benefits from more guidance than an experienced full-time data scientist would need to complete the work. But in the end, an intern experience is very similar to full-time work.
What university recruiting looks like at Microsoft
University recruiting at Microsoft is used to hire top talent from universities (including from undergraduate, master’s, and Ph.D. programs) to start careers at Microsoft. In the rest of this article, we outline the university recruiting process for data science.
The overall recruiting process consists of the following stages:
Application
The first step in the process of being considered for a data science internship or full-time opportunities at Microsoft is to apply online. On our careers website most of our positions are broken out into two sections, “Experienced professionals” and “Students and recent graduates.” Whether you are a university graduate, a soon-to-be graduate, or an intern, we have curated positions specifically for you in the “Students and recent graduates” section (see links included at the end of this article).
As you review the positions listed, you will notice that they are written broadly, and this is intentional. Our student and recent graduate positions encompass large numbers of open opportunities. So, when you apply to “Data & Applied Sciences: Full Time Opportunities for University Graduates,” for example, you are applying to all open positions across the company for university graduates in data and applied sciences. This allows you to be considered for multiple teams and opportunities with one application.
All applications, for both internship and full-time positions, are considered on a rolling basis. Although many of our interviews take place between September and December, we recommend that you apply when you feel ready. If you believe you would be more successful after completing another quarter or semester, feel free to apply between December and February. Most of our interviews for internships conclude in February and for full time positions around March, though there may be exceptions.
When you apply, make sure to submit your most up-to-date résumé or CV. Within your résumé or CV, highlight your strongest projects, research, and previous work experience in internships or previous full-time employment by clearly indicating the problem you were solving, how you solved it, and the end result, while also adding any technologies you used. This will give our recruiters and hiring teams the best understanding of your previous experience and areas of expertise. During the interview stage, hiring teams often rely on your résumé or CV when building out and aligning interview questions.
After you have submitted your application, someone from our recruiting or hiring teams will review it and determine whether we have an opportunity that aligns with your background. If we do, you will be contacted about the next step, a phone interview.
Phone interview (or technical screen)
The first-round interview typically lasts from 30 to 45 minutes and is conducted through Microsoft Teams or over the phone; sometimes interviewers will also screen share for coding problems. Phone interviews consist of both behavioral and technical questions, so be prepared to answer questions or solve problems using any technology listed on your résumé or CV. This conversation is a two-way street, so make sure to also have a couple of questions to ask your interviewer as well.
Final round interview
After the initial interview, you may be invited to the next step in the process, which consists of final-round interviews. As with our first-round screens, our final-round interviews are also conducted over Teams, and so screen sharing is quite common. Typically, you will have three to five interviews. Each of these interviews lasts about 45 minutes, with 10 to 15 minutes in between for you to take a short break and gather your thoughts.
Ph.D. presentation: If you are a Ph.D. candidate interviewing for a full-time data scientist position, you may be asked to do a one-hour presentation prior to your one-on-one interviews. The presentation consists of two parts: the presentation and a Q&A session. The format of the presentation is meant to be conversational and center on the focus of your Ph.D. work. The presentation sets the foundation for understanding your research prior to your one-on-one interviews.
Data Science interviews focus on four main components: technical excellence, collaboration, drive for results, and adaptability. We suggest brushing up on your math and statistics foundations, Machine Learning fundamentals, algorithm design, and data analysis skills, as well as coding for the technical portion of the interviews. Additionally, if you specialize in a specific domain or have in-depth knowledge in a specific topic area, be prepared to answer deep technical questions associated with it.
Beyond the technical aspect of the interview, we are also evaluating your ability to collaborate, drive for results, and adapt. During the collaboration portion of the final-round interviews we are looking for examples of how you have effectively worked with others to solve problems and build solutions. With regard to the drive for results section, the questions will focus on end-to-end problem solving while driving to positive outcomes. Finally, adaptability relates to the way in which you respond to changes in the scenarios and so questions will be asked on how you work under ambiguous conditions. Your interviewers are there to guide you through your interview and get a better understanding of your fit within data science at Microsoft.
Offer
Following the final-round interviews you can expect to hear back from your recruiting point of contact within about two weeks. If you are successful, your recruiter will notify you of the good news and share an offer letter with you. The offer letter will cover all benefits and compensation details. Your recruiter is there to be your advocate and liaison between the business and you. In addition to your recruiter, you will also be in contact with someone from the business group you received the offer from. They will be a great point of contact to ask any content and team-related questions you might have.
If you accept the offer, you will work closely with our onboarding team to ensure everything is in place for a smooth start. When you begin your career with Microsoft, we empower you through the Microsoft Aspire Experience. Those hired are invited to participate in this two-year learning and development experience where you’ll build your network, cultivate intentional capabilities, and gain perspective into the career opportunities across Microsoft’s many exciting businesses. Beyond the Aspire Experience, your team and manager are there to help set you up for success with Microsoft by providing training resources and support.
We are excited to see your application and consider you for data and applied science positions at Microsoft!
Check out the following links to apply for the 2022 season:
