A Day within the Lifetime of a Senior Data Scientist



Including a rundown of a standard step-by-step project outline

Photo by Slidebean on Unsplash [1].
  1. Introduction
  2. Responding, Planning, and Meeting Updates
  3. Expected Data Science Work
  4. Summary and Personal Thoughts
  5. References

Target market:

This text is meant for many who are current data scientists and keen on becoming a senior data scientist. It may well also serve for example of working in data science normally for those trying to switch careers.

Let’s get the very first thing clear, a senior role is different from company to company, so please take this day-in-the-life with a grain of salt. The principal difference between a senior and non-senior role, to me anyway, is that a senior position normally involves more collaboration cross-functionally between different teams and stakeholders. It can also be the case that you just own a complete project, very much like a product manager, where you will not be only coming up with the answer, but additionally coming up with the why and its impact, say KPIs (Key Performance Indicator), for instance. With that being said, let’s jump into an example of a typical day for a senior data scientist.

Photo by Alvaro Reyes on Unsplash [2].

Normally within the morning

I’ll attempt to order these principal events so as of time, but be mindful these can flip depending on the demand of the duty at hand.


You could be surprised to seek out out that quite a lot of your mornings or day for that matter, may be composed of responding to Slack threads. These discussions can range from anything like an easy clarification of information science jargon, to a long-detailed thread response backwards and forwards between multiple those who is more-or-less a brainstorming session on the subsequent steps for a selected project.

Slack or whatever tool your organization may use generally is a pro or con as it may possibly be distracting to some at different times, but it may possibly even be a quicker solution to answer questions than participating in a complete meeting, whether that be in person or on a video call like Zoom.


Chances are you’ll also not have any urgent notifications to reply to within the morning, but either way, it would be best to plan out the remainder of your day based on the priority of the asks. Sometimes your duties will actually be ranked based on the priority level, so you’ll have a transparent expectation of the order wherein you will want to execute your tasks, whether that be an information request, or an extended project like obtaining features to your data science model. These asks can actually be a mixture of others requesting from you to yourself organizing your personal requirements for a project.

Meeting Updates — Standup (as some corporations call it)

Next, you’ll normally have some sort of morning meeting where you’ll update a stakeholder(s), manager, and other people in involved in a project. These meetings are vital in order that everyone seems to be on the identical page about what has been done thus far, and what still needs work, and they sometimes are available the shape of a small, quick meeting time so it’s as efficient as possible, not taking away time from others’ day.

Now that we’ve got a general concept of a typical morning, let’s give some specific examples of those events:


  • Answer clarifying, urgent Slack questions which can be normally not needed to be an official task ticket — ex: “do you mind sending me the dashboard link for this evaluation”, “are you able to explain to me what MAE means?”, and “what are the principal KPIs for this AB Test?”, etc.


  • Based in your responses, you may create an official ticket (a well-liked project management tool, Jira, for instance, which uses a Kanban board) — ex: there can have been a disagreement between the KPIs for a project based on disadvantages that arise, so a ticket might be created to investigate one other KPI that is maybe faster and achieves similar, useful decision-making criteria

Meeting Updates:

  • Updating stakeholders on what happened, what went fallacious (if anything did), what’s the blocker (if there may be one), what is required, and what are the subsequent steps — and who else must be pulled in to reply any outstanding questions — ex: we’re blocked on this project because there may be an excellent ticket to include a latest column right into a database table that is required for the info science model feature set
Photo by Jefferson Santos on Unsplash [3].

Updating Ticketing

Assuming you might be using some kind of ticketing system to arrange tasks, just like the aforementioned Jira tool, it would be best to update what’s in your backlog, in progress, and what’s now done. These facets could be different in your organization. For instance, you may have a planning section too. Crucial part is that your team is aligned on what each bucket means, whatever the title of the bucket and what number of you have got.

On this section, I can go a bit more into detail on data-science-specific tasks, while the seniority aspect of it plays the a part of end-to-end project organization.

Project Outline:

Depending on the day, you may perform the next, one+ tasks, with an example of every:

  • Find a chance — ex: chances are you’ll notice a certain category of your organization’s product has low sales
  • Define the issue concisely — ex: “the pants category has the bottom sales out of any category”
  • Asses possible impact — ex: “pants make up 80% of our inventory, yet it’s our worst performing category when it comes to sales”
  • Develop an answer (data science or mixture of) — ex: the pants category was incorrectly categorized since it didn’t discern between shorts and pants properly, so the inventory was incorrectly classified in ‘all-other’, and other people didn’t expand that category on the homepage. This incorrect classification was because the corporate was using a homemade rules-based solution. The answer is a choice tree classifier categorization.
  • Map out what is required from others — ex: you will want certain data like an outline column in a database table to make use of as a model feature that can help the model classify accurately, so that you will work with the info engineer to provide you with a means of ingesting product data
  • Test the answer locally/dev environment — ex: create an end-to-end means of ingesting a dataset, training and testing the model algorithm, and deploying the model endpoint in order that it may possibly be utilized by a service that routinely classifies any latest, incoming inventory, in addition to reclassifying ad-hoc, incorrectly categorized items. You will even be certain that the answer is definitely higher by proving it with real data, and that the machine learning operations process works as well.
  • Test the answer based on KPIs in production — ex: you’ll be able to AB Test your data science solution using KPIs like ‘pants sales’ for instance, to see if the expected impact actually occurred. Chances are you’ll work with a separate AB Testing specialist at your organization or perform this yourself.
  • Assess the actual impact — ex: relay your findings from the test to stakeholders in order that everyone seems to be aware of the answer’s results.
  • Go forward (or not) with the answer! — ex: depending on in case your test was successful, you’ll execute your production-ready solution to all the items moving forward. If the expected final result didn’t occur, you can iterate on the answer based on those findings appropriately — possibly you overfitted your model, or must train with more data so your model generalizes higher, for instance.

These principal bolded points normally can account for many of the process you’ll see on a day-to-day basis depending on where you might be at along with your project. You may additionally be working on one+ project at a time. You may additionally be more involved with working with more stakeholders and executives at your organization in a senior role.

Every company has its differences and each role has its differences. Nevertheless, there are still commonalities between senior data science roles. To summarize, the principal takeaway is that senior data science positions do all that a standard data science position entails, but in addition they deal with the project from end-to-end more and act as a product manager for that project, entailing more responsibility in the method as an entire.

What I learned from being a senior data scientist?

The principal things I learned as a senior data scientist are the importance of:

  • Communication
  • Prioritization
  • Willingness to pivot

Within the senior position, you’ll have more interaction with product managers and company executives, so with the ability to communicate clearly and efficiently is a must. Next, prioritizing your tasks and principal data science projects is incredibly vital because you will see that that there may be this unlimited amount of labor you’ll be able to do, nonetheless, it is vital to grasp how one task or project compares to a different. Lastly, at the same time as you propose your roadmap, you’ll have for use to pivoting to a special project as priority and urgency can change around a certain topic. These three facets of the role are especially prevalent within the senior position since you not only serve because the subject-matter expert, but you furthermore mght are likely to serve because the leader of the info science space normally in your organization as well.

What surprised me and what motivates me?

The most important thing that surprised me concerning the senior position is that you’re going to not be doing as much data science as you’ll expect. As a substitute, there are other parts of the business which can be just as vital, like what I highlighted above that may take quite a little bit of trip of your day.

What motivates me is seeing the impact the senior position has on the business, as making a case for a certain data science project backed by a data-driven strategy and leading it from end to finish, is each vital and empowering.

The main target in additional on the strategy, and here’s what we discussed:

Responding, Planning, and Meeting Updates
Expected Data Science Work

I hope you found my article each interesting and useful. Please be at liberty to comment down below in case your experiences are the identical or different as a senior data scientist. Why or why not? What other things do you think that ought to be discussed more, including more pros and cons? These can actually be clarified even further, but I hope I used to be capable of shed some light on what to anticipate on this position.

I’m not affiliated with any of those corporations.

Please be at liberty to ascertain out my profile, Matt Przybyla, and other articles, in addition to subscribe to receive email notifications for my blogs by following the link below, or by clicking on the subscribe icon on the highest of the screen by the follow icon, and reach out to me on LinkedIn if you have got any questions or comments.

Subscribe link: https://datascience2.medium.com/subscribe

Referral link: https://datascience2.medium.com/membership

(I’ll receive a commission for those who enroll for a membership on Medium)

[1] Photo by Slidebean on Unsplash, (2020)

[2] Photo by Alvaro Reyes on Unsplash, (2018)

[3] Photo by Jefferson Santos on Unsplash, (2017)


What are your thoughts on this topic?
Let us know in the comments below.


0 0 votes
Article Rating
1 Comment
Newest Most Voted
Inline Feedbacks
View all comments

Share this article

Recent posts

Would love your thoughts, please comment.x