By Austin Gorsuch

There is the world of business and the world of data science, each with its own processes and priorities, its own recurring problems and hard-won solutions. But as these worlds continue to merge and evolve with each other in our data-driven economy, new roles emerge to act as a bridge between them, to balance the needs of one world with the needs of the other and produce outcomes that work for both parts of the data-driven business equation.

One such role is the data science product manager (DSPM). In this article, we’ll break down the roles and responsibilities of a DSPM, how you can go about becoming one, the sorts of interview questions you can expect to face if you’re pursuing it as a career, and your prospects as an data science product manager professional.

What is a data science product manager?

In simple terms, a data science product manager acts as a liaison between a business and a data science team to produce and maintain one or more data science products.

What is a data science product? It might be a machine learning algorithm that is meant to be deployed across vast amounts of data, or a fully integrated suite of dashboards meant to keep the business informed about its daily function, or any number of alternatives.

As a DSPM, not only do you have a strong understanding of machine learning and the needs of your engineers, but also the ability to develop business cases for complex and changing technologies and communicate with stakeholders about the product in development. A DSPM is involved in the evolution of a data science product from its inception to well after its initial deployment, serving as a crucial source of input in its maintenance and continued development.

Some of the day-to-day roles and responsibilities of a data science product manager are:

  • Managing a product that involves the deployment of machine learning models
  • Analyzing data to influence product decisions
  • Balancing trade-offs between different possible machine learning algorithms
  • Communicating between a team of data scientists, machine learning engineers, and external stakeholders
  • Prioritizing next steps in the data science product lifecycle

It’s important to note that a data science product manager, who manages a data science product, is distinct from a data product manager, who manages data as a product.

While there are some similarities between the two roles, such as acting as a bridge between internal stakeholders and a data science team, a DSPM must have a deeper understanding of the actual technical process of developing a data science product, rather than merely dealing with the outcome (the data) that such a product makes available to a client.

How do you become a data science product manager?

Data Science Product Manager Career Path
Image from Pixabay

Case I: The Data Scientist Turned DSPM

Sam has worked as a data scientist for five years now and he just can’t seem to get on the same page as everyone else. He’s less concerned with refining the same old model to make it one percent more accurate than he is in making sure the algorithm is useful to the customer. He doesn’t have anything against his coworkers, he just doesn’t experience the joy of data science for data science’s sake.

So he resolves to make a change in his career. He takes a few business classes at night and takes an online course on product sense. Then, he puts out his updated resume on LinkedIn and Indeed and soon he’s interviewing for the role of data science product manager. He struggles in his interviews at first, but he practices for the questions he struggles to answer the most: questions on product sense and business case.

Within a year, he lands a new job as a data science product manager at a start-up specializing in providing machine learning solutions in a B2B context.

Case II: The Product Manager Turned DSPM

Alice has worked as a product manager since she graduated from college with a Masters in Business Administration. She’s job-hopped from company to company and has developed a reputation for being able to take any product and turn it into the company’s flagship. The only problem is that Alice feels like she’s learned everything she’s going to learn and is bored in her current role.

Alice has recently found herself gravitating toward data science as someone interested in future-facing products with durative value for the customer. She wants to be at the head of a data science engineering team charged with making deep learning algorithms that help chart a course for businesses with innovative models and the customer at the forefront.

So she resolves to make a change in her career. She begins with an online course on machine learning, then goes back to school and immerses herself in the world of data science with a specialization in natural language processing.

Afterwards, she puts her updated resume out to the world and finds that many companies are looking for a candidate with her new skill set. Her job-hopping experience, along with some practice for the machine learning parts of the interview, has made Alice a pro at the interview process and she wastes no time in finding the right job for her.

Now Alice works as a data science product manager for a company that builds recommendation algorithms for various B2C streaming companies looking to give their consumers a totally new kind of viewing experience.

If you, like Sam or Alice, are looking to take an online course to help prepare you for a career change to data science product manager, check out our online courses on product sense, machine learning, and more!

The Data Science Product Manager Interview  

If, like Sam and Alice, you’re looking to be a data science product manager, here are some questions that you might encounter during the interviewing process. We’ll look at questions concerning product sense, SQL, and machine learning:

Product Sense: Green Dot

Let's say we're working on a new feature for LinkedIn chat. We want to implement a green dot to show an “active user” but given engineering constraints, we can't AB test it before release.

How would you analyze the effectiveness of this new feature?

Watch this video for the full solution:

Test your skills on this question on product sense on Interview Query.

SQL: Employee Salaries (ETL Error)

'employees' table

+---------------+---------+
| column        | type    |
+---------------+---------+
| id            | integer |
| first_name    | string  |
| last_name     | string  |
| salary        | integer |
| department_id | integer |
+---------------+---------+

Let’s say we have a table representing a company payroll schema.

Due to an ETL error, the employees table instead of updating the salaries every year when doing compensation adjustments, did an insert instead. The head of HR still needs the current salary of each employee.

Write a query to get the current salary for each employee.

Assume no duplicate combination of first and last names. (I.E. No two John Smiths)

Here’s a hint:

The first step we need to do would be to remove duplicates and retain the current salary for each user.

Given we know there aren't any duplicate first and last name combinations, we can remove duplicates from the employees table by running a GROUP BY on two fields, the first and last name. This allows us to then get a unique combinational value between the two fields.

Try out this question on our SQL editor on Interview Query.

Machine Learning: Job Recommendation

Let's say that you're working on a job recommendation engine. You have access to all user LinkedIn profiles, a list of jobs each user applied to, and answers to questions that the user filled in about their job search.

Using this information, how would you build a job recommendation feed?

Here’s a hint:

What would the job recommendation workflow look like? Can we lay out the steps the user takes in the actual recommendation of jobs that allows us to understand what a potential dataset would first look like?

Try this question on machine learning for yourself on Interview Query.

What do a data science product manager’s prospects look like?

A data science product manager can look forward to career opportunities at companies like:

  • Amazon
  • Netflix
  • Square
  • Google
  • Apple
  • Uber
  • Microsoft
  • eBay
  • LinkedIn
  • and many more!

Furthermore, as of 2021, Glassdoor lists the range of salaries for the DSPM position as $71,000-$146,000, with an average salary of $109,000. As you might expect, this means that the position is in high demand, not only among employers but also job candidates as well.

Luckily, Interview Query has a number of courses on product sense, machine learning, as well as statistics and A/B testing, that can be of use to an aspiring data science product manager. If you’re interested in the DSPM role, or ready to make a change in your own career, take every opportunity to get the edge on your competitors in the interviewing process.

In addition to our courses, we also host a number of real interview questions from companies like Google, Amazon, Netflix, and more, for you to practice on to make sure that you’re as prepared as possible for the interview that could land you your dream job. Don’t hesitate; sign up for your free account now!