Previously, we discussed the various roles in data science (data scientist, applied scientist, research scientist, machine learning engineer) and read about a data scientist who found himself mainly doing non-data science, SQL monkey work.

Now, you might be thinking, how can we prevent ourselves from getting into a role that’s not what we expect? Or if we’re in such a situation, what can we do?

How not to get into the wrong role

The first line of defense is to read the job description carefully. Don’t just focus on the job title! In this (slightly dated) study, researchers found that job seekers mostly paid attention to the job title and company, neglecting other information such as job requirements. We should spend time examining the requirements and job scope too. Most of the time, hiring managers and recruiters make an effort to customize these to match the role.

If we land an interview, we should ask the right questions. It’s as much about us assessing them as them assessing us. Here are some useful questions to suss out the role:

• What’s my day-to-day like: This helps with understanding the actual tasks done
• What tech-stack will I use most often: If you expected Python and deep learning libraries (e.g., PyTorch), but the actual work is more of SQL, dashboarding, and scikit-learn, then you might want to adjust your expectations or move on.
• What skills are needed to be successful: If you expected to guide decisions with analysis and storytelling, but they need software engineering and devops to put models into production, it might not be a good fit
• What are my expected deliverables: Check if it matches your expected balance of writing code, writing documents, doing presentations, etc.

More about the goals, skills, tools, and deliverables of the various roles here.

If you’re fairly certain about the role (and maybe have an offer), contact future team members you’ve met—during the interview cycle—to learn more. Or look them up via LinkedIn. In addition to the questions above, ask about the culture and your future boss. Usually, peers don’t have as much of an incentive to “sell” you the role and tend to provide more objective information.

We can also read reviews on sites such as Glassdoor. This can be useful for organizations with a single, centralized, data team. Nonetheless, note that in large organizations with multiple data teams, each team might be structured differently and have different cultures. Use your judgment and take these reviews with a pinch of salt.

What if I have the right role, but the wrong title?

For example, you might be building machine learning models, putting them into production, and working on engineering and devops tasks. However, you have the title of data scientist while peers (in other organizations) have titles such as applied scientist or machine learning engineer.

Depending on how you define data scientist, your title may not reflect the work you do. But I think it’s not that bad. At least you’re on the right side of the problem; it’s way better than having the right title but work on the wrong tasks (like this guy).

Don’t let your title define you. (Your job should not define you either for that matter.) If you’re doing the right work and growing towards your aspirations, just keep at it—this is what matters. In interviews, people care about the work you’ve done and what you can do, not your title.

What if I have the wrong role?

For example, you might have joined a team expecting to build and deploy ML models. Instead, you find most of your time spent on foundational data engineering work and providing analysis for business decisions.

First, assess if the situation is temporary or permanent. If we joined a start-up with nascent data capabilities, the initial data engineering and ad hoc analysis might be unavoidable (and hopefully, temporary). On the other hand, if we joined Facebook as a (non-core) data scientist, the core work of data extraction, analysis, and statistics is likely to be permanent.

Assuming it’s only temporary, I think it’s healthy to embrace it and enjoy the journey. Hey, at least we get to work with data. It also stretches us outside of our comfort zone, on tasks we don’t usually work on (e.g., data engineering, infra)—this is a great learning opportunity. (The situation could be far worse; imagine being conned into a company that does not have data…)

At the same time, work with your boss to carve out time for a 20%/research project. It should match organizational needs and your aspirations. Such projects can directly improve revenue, cost, or customer experience, or build team capabilities (e.g., internal library for rapid experimentation). If the project goes well, you could be spearheading a new initiative and get to write your ideal job description.

What if there’s zero opportunity to work on such projects? Well, it’s not the end. Most of us work 40 - 50 hours a week; there’s plenty of personal time for self-learning and projects. And public data is widely available. Personal projects are a great way to learn new techniques and gain hands-on practice. It’s also easier to share the code and write about our process, making it a solid addition to our portfolios.

What if I change jobs (same role), but get a lesser title?

This is similar to the situation of right role, wrong title. Having a less prestigious title can hurt the ego. Nonetheless, try to push through and focus on the learning and impact of the new job. Let your growth, work, and results speak for you; don’t use the title as a crutch.

Depending on the new company, you can try negotiating the job title with your boss and HR. Success largely depends on the existing organizational structure. If the title is fairly standard (e.g., data scientist) and several others have the same title, it might be a hard sell. (The hiring manager might have a tough time explaining the title difference to existing team members, or have to rebrand everyone). But if the role is fairly new and HR is flexible, changing your title might be possible.

If you need to invoke your academic pedigree or job title for people to believe what you say, then you need a better argument. - Neil deGrasse Tyson

Prevention is better than cure

Do the due diligence before accepting the job offer; this is the best way to prevent a role or title mismatch. Ask the right questions, contact future teammates, read the job reviews.

If you find yourself in the wrong role, assess if it’s temporary and how much is within your control. Focus on what you can control (it’s healthier this way). Reframe your mindset, find opportunities for 20% time projects, and do self-learning and projects in your free time.

Overall, if you’re doing meaningful, high impact work, learning lots, and have great peers, don’t let the title weigh you down. It’s just a title—your competencies, results, and job satisfaction matter much more.

Thanks to Yang Xinyi for reading drafts of this.

If you found this useful, please cite this write-up as:

Yan, Ziyou. (Nov 2020). How to Prevent or Deal with a Data Science Role or Title Mismatch. eugeneyan.com. https://eugeneyan.com/writing/role-title-mismatch/.

or

@article{yan2020mismatch,
  title   = {How to Prevent or Deal with a Data Science Role or Title Mismatch},
  author  = {Yan, Ziyou},
  journal = {eugeneyan.com},
  year    = {2020},
  month   = {Nov},
  url     = {https://eugeneyan.com/writing/role-title-mismatch/}
}