4 Realistic Career Options for Data Scientists

It’s almost 10 years since "Data Science" became mainstream. We ask less about how to get into Data Science, but wonder "what’s next?" This article includes insights on four non-trivial, but practical, options and their pitfalls.

By Ian Xiao, Marketing & AI Practices at Deloitte.

Photo by Caleb Jones on Unsplash.


February 27th, 2020


About 3 weeks before the city was locked down due to COVID-19, a friend and I sit at our favourite Thai restaurant in downtown Toronto.

“Should I stay in Data Science? If not, what should I do next?” My friend asked.

Like the sound of gathering and smell of fresh restaurant food, the question of how to get into data science seems a bit distant now. In fact, you can find ~3 billion results on Google with a ridiculously specific step-wise guide (only ~2 billion on “how to make money”).

Google Search as of March 25th, 2020.

Yet, my friend’s question lingered. Deep down, the tension is driven by two factors: 1) expectation vs. reality of data science — it can be more boring than we expected, and 2) the role vs. our aspiration.

So, what are the options?

  1. Stay in the same role and internalize the frustration (you can skip this article, but please share before clicking off ????)
  2. Leave the current role, but keep doing Data Science-related stuff elsewhere (this article is for you)
  3. I am totally done (are you sure? think about option 2 before landing on this)


But, where to?


To know where we want to go, you must know where we are. To get a bigger picture of the alternatives, many of us may turn to the question of “what is data science?” If you Google this phrase, we will likely come across something like this:

What is Data Science, Image Source.

This categorization makes sense, but it doesn’t really capture the reality and nuances of the day-to-day work and the nature of the company. These two factors are arguably the most important when we make professional decisions.

There is a better way of thinking about this. So, let me present to you the Archetypes of Data Scientists.


Just a Second...


Before we start, it is important to consider my background and journey, so you can translate my insights based on your own reality and constraints.

In short, I witnessed the evolution of Data Science by consulting for large enterprises, befriending many types of Data Scientists, and building ML products at an AI start-up that ended up being acquired.

My opinion is only one data point (hopefully it’s a useful and unique one), so keep your eyes out for others’.


Options: Archetypes of Data Scientist


So, what is Data Science? It depends on the size of companies (enterprise or start-up) and the primary responsibility of the role (client-facing or internal focused).

When you interview for a data science role, despite the title and job description, it’s likely to fall into one of the following four oversimplified and subjectively labelled groups.

Four Archetypes of Data Scientists. Author’s Analysis.

How can you use this? The Archetype shows you the possibilities. It allows you to see where you are right now and what the immediate options are. For aspiring data scientists, you can use this to figure your best starting point.

Next, you might ask: how do I know if the new role is better for me?

To help you decide, here are some pros and cons of each role. In addition, I will discuss the talent profile that might fit the best. Of course, these are somewhat subjective: what’s good for me might be bad for you. There are always exceptions. So make your own judgement.


  1. you can also categorize by other aspects of the company, such as industry and type of product (e.g., data science-related services or tools, non-data science products). I think the size of the company has more impact on the day-to-day work.
  2. company size isn’t binary, there are many medium-sized companies; I binarize it for simplicity.
  3. at start-ups, the wall between client-facing and internal stuff doesn’t really exist; people tend to work on both. It’s just a matter of time allocation.


1 — The Expensive Consultants


Data Science service at global consulting firms or professional service arm of big technology firms (e.g., Deloitte, McKinsey, Accenture, Google, IBM, etc.)


The Good: Very good training on problem-solving and how to work productively. Work on very “important stuff” because clients pay the high fee for results. Every project can be different in case you are bored or frustrated. Very clear promotional path and requirements (e.g., Analyst, Manager, to Partner). Exposure to senior executives, a wide range of topics in early careers, and lots of ambitious people.

The Bad: Business value out weights everything (e.g., scientific innovation and cool algorithms). Demanding hours across all seniority. Some firms may not consider Data Scientists for the traditional Partnership track (you might feel sidelined, but don’t worry, most people are very respectful). Lots of alpha personalities.

Best Fit: Business-oriented people with some technical training and aspiration to run your own “small business” within a big company. Students who are just starting out, looking for mentors, and want to learn the best lingos and don’t mind being worked really hard.


2 — The Tech-Savvy Sales


Technical sales or project lead roles at AI-focused start-ups or medium-sized companies (e.g., DessaElement AIH2OClouderaPalantir).


The Good: You can work on some cutting edge use cases because clients typically look to you for innovation, instead of delivering (boring) long term, large scale transformation projects. A bigger voice on important strategic and product decisions. Be nimble and innovative.

The Bad: Some clients may not trust you with large projects (e.g., get more budget), which is the flip side of being “nimble & innovative.” May need to do lots of “free” work to win client trust. Less back-office support compared to bigger firms. Product vision can be influenced by investors (if you have the wrong VCs) or the sunk cost mentality, instead of true market demand.

Best Fit: People who want to grow with the company in the early-stage, live the entrepreneurial dream, and already have a solid business network, domain expertise, and/or reputation. Not good for newbies because you need to hit the ground running.


3 — The Product Hippies


People with Software and/or ML background who work with Engineering to build IP, demos, and support sales calls at AI product companies.


The Good: Work on interesting and practical problems without having to deal with too much client politics. Shorter-term projects. Get to influence or define how the products should be designed as an “internal user.” Build cool stuff for the clients or internal.

The Bad: Likely to get pulled into a client- or user-facing role, which creates tension in competing priorities and time management. Hard to find the right balance between “cutting edge” vs. “immediately practical.” Might get sucked into doing endless customer support work.

Best Fit: People who are product-oriented, engineering-focused, but some times hacky. Seasoned professionals with domain expertise in certain technology stack or workflow. Students who have an open and curious mind, enjoy technical challenges, and can get sh*t done.


4 — The Big Family


Okay, families can be complicated, so this group needs to be broken into two segments: the Unsung Heros and the Super Geeks.

The Family Structure. Author’s analysis.

4.1 — The Unsung Heros

People from traditional BI, Analytics, and Modelling groups within enterprises. They mostly work with lines of businesses or functions (e.g., Marketing, Risk, Finance, etc.) They were the Data Scientists before the term was coined.


The Good: Very focused work. Close to real business operations. Have access to unique datasets. Have access to operationalized and large scale infrastructures. Great work-life balance. Have influence or power to make investment decisions.

The Bad: Things can be slow and boring at (most) times. Often don’t get the recognition of having the sexiest job of the 21st century. Internal politics. Slow(er) career trajectory. Get locked into certain roles or projects.

Best Fit: People who found their passions in life or enjoy investing time in specific domains. People who don’t care about the hype. People who have lots of patience and resilience.

4.2 — The Super Geeks

The stereotype of the “golden child” in Data Science. People who work at R&D in major companies, such as Google Brain / DeepMind, Facebook’s FAIR, Uber, and Walmart Research, etc.


The Good: Work on very intellectual topics. Have access to unique datasets and problems. Be able to push the envelope with the resources of the big firms. Get lots of recognition and praise.

The Bad: Strong tension from having to demonstrate “business value.” Business priorities can limit or influence research topics. Might get suck into “boring” implementation if the research demonstrated value.

Best Fit: Academics. Engineers with a very strong taste for academic research and educational background. Graduate students who are looking to publish and get some industry exposure.


What’s Next


Now you have a good grasp of what’s possible, ask yourself these questions:

  • Where am I now? Why am I frustrated?
  • What do I really want in my data science career (or life in general)?
  • Does the destination offer perceived relief and benefits? (Hope this article provided some insights)
  • What are my attainable options?
  • How do I get there with the least effort and the best outcome?

Original. Reposted with permission.


Bio: Ian Xiao helps clients transform core businesses or launch new capabilities with digital technology and practices (e.g., AI, Big Data, Agile, Design Thinking), and advises B2B data and AI start-ups. Ian is a Top Writer in Business & Tech on Medium along with MIT Review, Fast Company, and Google.