A Day in the Life of a Data Scientist: Part 4
Interested in what a data scientist does on a typical day of work? Each data science role may be different, but these contributors have insight to help those interested in figuring out what a day in the life of a data scientist actually looks like.
Sure, no data scientist role is the same, and that's the reason for the inquiry. So many potential data scientists are interested in knowing what it is that those on the other side keep themselves busy with all day, and so I thought that having a few connections provide their insight might be a useful endeavor.
What follows is yet another round of some of the great feedback I received via email and LinkedIn messages from those who were interested in providing a few paragraphs on their daily professional tasks. The short daily summaries are presented in full and without edits, allowing the quotes to speak for themselves.
If you are interested in the first 3 posts of this series, you can find them here:
- A Day in the Life of a Data Scientist (Part 1)
- Another Day in the Life of a Data Scientist (Part 2)
- Yet Another Day in the Life of a Data Scientist (Part 3)
Rogelio Cuevas is a Data Scientist at the Data Science and Model Innovation team at Scotiabank, Toronto.
Being a data scientist involves much more than the technical side of the job. As data scientists we are expected to solve complex business problems and make sure that the models we develop provide value to the business. For this, it is crucial to understand what is important to the business and stay in constant contact with business units to avoid divergence between what we develop and what they need.
I cannot say there is a typical day in the job but I can say there are three main components that are present in every day; the only changes are the proportions of these components that constitute every day: meetings, technical and administrative.
My day starts around 9:00 A.M. with a 15-minute long Agile stand-up meeting. I meet with people involved in the projects I work on and provide 3 pieces of information:
- What I did the previous day,
- What I will focus on during the day, and
- Any blockers I may have and if I need help from other team members to make progress.
In addition to these meetings there are many others that I need to attend to cover various needs. Some of them have the goal of updating business units and internal clients with findings and progress I have made, request their input or establish priorities in future tasks.
Other meetings are needed to get clarification or provide information to understand future deployment of models we are currently developing or to ensure that our understanding of the process is indeed correct.
Regardless of the details of the meetings they all have a common goal: satisfy the business needs and ensure we are in total alignment with them.
Developing models that satisfy business needs is an iterative process that involves
Gathering and cleaning data,
- Creating a dataset based on which we will develop a model,
- Defining a target variable that satisfy business needs,
- Generating preliminary models and improving their accuracy, and
- Transforming the accuracy of the model into metrics that are relevant for the business.
We use various tools and algorithms:
- SQL for gathering and grouping data under a single dataset,
- Python and R to cleaning and creating models,
- Git for versioning control of code we generate, and
- Various regression and classification algorithms to make predictions.
These tech tasks are collaborative and can go both ways: I can create a pull request and ask one of my colleagues to review it or I can act as a reviewer of a pull request created by a colleague. Either way, collaboration is key to make incremental improvements in the products we generate.
Given that data science is a field that changes very rapidly we are expected to learn as many tools and techniques as we can. We need to be flexible and open to constant learning.
I also devote a good portion of our time to administrative tasks such as
- Booking meetings
- Replying e-mails and having virtual calls
- Creating decks to report our findings
- Generating reports
- Documenting processes and models we are developing
Outside of office hours I try to be as active as I can by attending Meetups I find interesting and relevant, networking with other data scientists, and reading blogs, books, papers and packages documentation I come across.
It is a challenging and exciting role. I feel very fortunate of being part of a team that is highly collaborative and committed to push ourselves to learn as much as we can.
I hope you find this note informative and it gives you some understanding of what data science looks like today in the world of business.
Sarah Nooravi is a Senior Data Scientist at Operam, Inc., based in Los Angeles.
Keeping in mind that no two data science positions and no two data scientists are exactly the same, let me walk you through one of my days! I hope it helps you see that a data scientist’s day is not just filled with modeling and big data. There is a lot of learning and time for communication.
A typical day for me starts the moment I get in my car. With a cup of coffee in hand and a two hour commute (thank you, Los Angeles), I am (usually) excited to start learning and doing something productive. I use this time to catch up on my favorite podcasts (Masters of Scale, Recode Decode, or Data Skeptic), listen to YouTube videos on ML/DL, prep for my next teaching lesson, brainstorm new product ideas/approaches, or put on some good music to get me pumped for the day.
Once at work, I settle in with a cappuccino and a Spotify playlist. I then use the morning to get caught up on emails and Slack. The first thing I look for are any urgent requests from clients. If there isn’t anything on fire, I read the Business Insider updates that come in daily. These updates help me keep current on trends and new product features on the platforms with which we work. These updates are enormously helpful when it comes time to make high-impact decisions on campaign strategy.
The middle of the day could bring anything! Administrative requests, ad hoc reporting needs, internal meetings, client meetings, refinement of existing models, building out new tools, and many more things I cannot fully list here. Each of these tasks play a key role for the business and for my personal development. Administrative tasks like sprint planning and hiring challenges and improves my time-management skills and ability to make quick decisions. Client meetings challenges and improves my communication skills and business acumen. Refining and building out new models challenges and improves my technical skills and creativity. As a data scientist, I find cross-functional roles that allow you to develop holistically to be the most exciting and rewarding.
Once I have put in a full day at work, I keep thinking about and learning data science. Depending on the day of the week, I keep my evenings full with other activities like: teaching a bootcamp on Data Visualization, appearing on a panel for Data Science Office Hours with other passionate data scientists in the field, mentoring women in engineering through Society of Women Engineers, or hosting my own Machine Learning Meetups. If you can’t tell, I can’t get enough of this field.
To recap: I start my day around 7am and end up back home around 10-11pm. The days are long but I love what I do. The key to making each day a success is happiness. I try to incorporate things in my day that bring me joy and help me grow, individually and within my community.
Kate Strachnyi is a Data Visualization & Reporting Specialist at Deloitte, located in New York City.
No alarm needed... my day starts at 6am. One of my two kids will surely wake up by 6am at the latest. I’ve held this early morning schedule for close to 4 years now and I love it! I make a cup of coffee, sit down, and check my phone (guilty of doing this first thing in the morning). After reviewing my emails/ messages for work, as well as my personal messages, it’s time to start getting my older daughter (age 3) to her school. We pack her lunch together, pick out her outfit and then spend some time playing before I drop her off at 8am. The best part of my routine is that I work from home (it is super cool that my job provides this flexibility). If the weather is nice and the mood strikes, I go for a run (about 2-3 miles) before starting work.
9am – typically log into my work computer and check my calendar to see what calls I have for the day; then I start work. My responsibilities include working on dashboards, developing KPIs, ensuring data quality and most importantly providing data transparency for leadership. I’m there to help people connect the dots using data. Mainly I do all of this via my favorite data visualization tool – Tableau! I love the tool so much that I have a YouTube channel where I provide tutorials on how to use Tableau – Story by Data.
Around 12pm I have lunch (usually something I prepare that same day). Then continue working, touch base with my team members and make sure we are all on track to accomplish what we set out to accomplish for the day/ week. At 2pm I go and pick up my daughter from her school and bring her home (it’s walking distance and takes only a few minutes; but I love the energy that I come back with after this short walk). The rest of the afternoon is spent on work and thinking about how to automate some of the work using technology.
Working with data has always fascinated me; the ability to derive insights from especially messy and unstructured data is what gives me energy. My curiosity to learn more about the field led me to writing my book Journey to Data Scientist which provides a series of more than twenty interviews with leading data scientists and uncovers key insights around getting started in data science.
Alexander Wolf is a Data Scientist at Dataiku, based in New York City.
I first took an interest in Data Science when I attended Machine and Deep Learning classes at Dartmouth as a computer science major undergraduate; I thought the material was intriguing, exciting, and new. After realizing the potential of this field, I was certain this was what I wanted to get into. I also knew I wanted to work at a startup.
From there, I taught myself how to use some data science APIs properly through online resources. I had already developed a good mathematical and theoretical machine learning foundation from my classes at school, but still needed to learn many of the basics of data visualization and engineering, such as how to efficiently index Pandas Dataframes, create Sci-Kit learn pipelines, visualize data with Seaborn/MatPlotlib, clean data properly with lambda functions, and more. (Python libraries; R equivalents exist). I found Datacamp to be a great resource for this. Similarly, I thought the Sci-Kit Learn documentation was great resource for reinforcing my understanding of each algorithm, and how its hyper parameters affect each model.
In early 2017, I ended up getting a job at Dataiku in New York City. Dataiku is a high growth international startup that makes a platform for Data Science; the company was founded in Paris and raised a 28 million series B this past September. From day one, I knew I’d made the right decision in both field and career path. Not only does Dataiku have a phenomenal company startup culture, but everyday I also get to work on something new and exciting. No day is the same here, which makes my job a lot of more fun.
Since we sell a platform for data science, the whole data science team is pretty heavily integrated with most other departments of the company - from helping the sales/partnerships team with data science expertise throughout the sales process, to helping the marketing team create relevant and at times technical content for meetups, our blog, 3rd party events, and so on. When I am not working with other departments of our company, I typically am supporting our customers, or furthering my Dataiku platform skills by working on some of my own data science projects.
When we onboard new customers, many of them want us to be onsite for 1-3 days to teach their data engineers, analysts, and scientists how to leverage Dataiku to enable collaboration between their teams, easily put data pipelines/ models into production, and teach them how to use many more of the features our platform offers. I typically need to do this two to three times a month while traveling all over the country and being exposed to different data science processes and teams. It’s very interesting to observe how different industries and companies are organized and what they are doing with their data.
Many companies also purchase professional services from Dataiku, and I get to work with a lot of different firms on their data projects. I have been exposed to a wide range of data science toolbox stacks, use cases, teams, and organizational structures. I’ve worked on NLP projects, time series forecasting, database deduplication, visualization, deep learning, production pipelines, Spark pipelines, and much more with our customers. One of my favorite parts about working at Dataiku, is that I’m not confined to working within one industry, or using one tool stack; every week I’m working with different companies, learning new things, and helping them truly get the most out of their data with Dataiku. I’m lucky to be working at a startup like this, and I really look forward to seeing it continually improve and grow.
When I am not in the office or traveling to meet with customers, I enjoy spending my free time reading tech/ AI news, coding, working on data science projects, playing basketball, brainstorming startup ideas, hanging out with friends, and reading the latest deep learning research papers. I am able to stay updated with news/ tools in AI and Data Science by browsing my LinkedIn feed; I actually like it more than any other social media platform Some great people to follow I recommend - Matthew Mayo, Florian Douetteau, Andriy Burkov, Rudy Agovic, Gregory Renard. I typically get up around 8AM and then take the subway to downtown Manhattan to work. I get home around 7PM. I try to work out as much as possible also, since I am an ex-D1 basketball athlete.
- A Day in the Life of a Data Scientist
- Another Day in the Life of a Data Scientist
- Yet Another Day in the Life of a Data Scientist