Is data scientist the right job for you?
A note from a data science hiring manager.
Data scientist has become one of the highest demanded job. As many are seeking to become a data scientist, today, I have put together a short summary of being a data scientist.
What does a data scientist do as written in job descriptions?
- Work with large, complex data sets doing data gathering, processing and analyzing.
- Inform and influence key product and business decisions by developing data product and/or running strategic analysis and applying advanced analytical methods (experiments, statistical or machine learning models) as needed.
- Design core metrics to serve as north-stars for team efforts
- Work cross-functionally with effective communication and presentation skills
Types of data scientists
Many companies have data scientists fall into the following three buckets. The above description is applicable for all three types of data scientists.
- Product: Focused on building new features or products for customers. Main cross-functional partners include Product and Engineering
- Growth/Marketing: Focused on growing the user base through acquisition and retention/engagement tactics. Main cross-functional partners include Marketing and Product
- Research/Algorithms: Develop new and scalable statistical or machine-learning methodologies. Main cross-functional partners include Engineering and Product.
The day-to-day work
In general, we spend our time doing the following work on a daily basis.
- Understanding the problem: The most important part of our job is to understand the business problem at hand. We need to be able to answer the question “what are we trying to solve for?” and “how will my work add value to solving that problem?” before jumping into any data science project. This will further help build a clear and effective research plan.
- Data collection and processing: Believe it or not, this takes the majority of the time in completing data science projects. No data is perfect, period. We need to double check how the data was collected and learn common ways the team/company define and collect data (e.g., is daily active user counted at account level or people level?)
- Running analyses: Once you understand the problem and have processed the data, analyze it with the “appropriate” methodology. Note that finding and applying the right methodology is one of the most important parts of a data scientist’s job. Applying a complex machine learning methodology to show off your skills is NOT the right path.
- Communicating insights and recommendations: Effectively communicating the findings to influence key business decisions is very important to make impact from your project. Visualizing the data, telling a story through the insights, and sharing recommendations for next steps is the last step of a data science project.
A common misconception data scientists have is that they will be able to apply all the great techniques they learned as soon as they start the job. I warn you it’s not all sunshine and roses. Often we are spending the majority of our time in meetings with cross-functional partners or sitting in front of our computer trying to understand the data and cleaning it.
Also note that as a data scientist becomes more senior in their career, one may spend more time on understanding the broader business problem, breaking it down into smaller and actionable analysis plans, and communicating with cross-functional partners.
So what do I, as a data science manager, look for when hiring
- Technical expertise: As written above, finding the right methodology to solve for a business problem through data is a required skill. This is why data scientists go through 3–4 technical interviews ranging from data wrangling (SQL), programming (Python/R), experimentation and machine learning.
- A problem solver: I look for candidates who proactively solves a business problem through data. A data scientist needs to throw questions and actively search for methodologies. One who can only apply one type of solution to a problem or one who gives a textbook-like answer is often not the best candidate.
- A strategic partner: I look for candidates who have great communication skills and can succeed in collaborative work environments. After all, a data scientist does not sit in the corner and stares at the data all day long.
I want to clarify that this doesn’t mean that companies or managers are looking for an excellent generalist all the time. Bigger corporations like Google or Facebook typically find generalists in the interview stage and then go through a team matching phase after one passes the interview. However, smaller companies or startups might hire for a specific skill at certain times. For example, when I was building a data science team at a startup, at one time I needed someone who could scale ML algorithms and at another time I needed one who could build non-conventional experimentation framework. At those times, I had to pass on candidates who were great overall but lacked deep knowledge in specific areas.
So do you really want to become a data scientist? If you do, here’s my last piece of advice:
- Get your basics right! Instead of spending time learning the functions to run machine learning models on Python, spend time understanding the properties of the model.
- Read a lot! Read how others are solving problems. There are many resources today, including Medium, LinkedIn, Quora, company websites and others where you can read about how people are solving business problems through data.
- Meet people! Of course we cannot do this during COVID-19. But once it gets resolved, meet others to hear how they tackle problems. Go to company meetups where other fellow data scientists share their experience.
Enjoy your journey as a data scientist!
