A data scientist is a cutting-edge professional responsible for developing AI models that drive various activities in our daily lives—whether it's recommending songs we might enjoy, predicting products we're likely to purchase, or forecasting a company’s sales.

Today, Sertis would like to introduce you to Earn, one of our talented Data Scientists. Earn plays a crucial role at Sertis, developing models that meet our client's needs and being part of a team that continually researches and innovates to advance modeling techniques.

Let’s take a look at what Earn does on a typical day, the tools she uses for her work, and what sparked her passion for this career.

What are the daily tasks of a data scientist?

“The daily tasks of a Data Scientist can be divided into three main categories: Pre-sales, Implementation, and Internal work.

Pre-sales is the stage where we pitch projects to clients. During this phase, we collaborate closely with the Business Development team and the clients to understand their requirements, brainstorm, and design solutions tailored to their needs. This often involves researching and conducting a literature review to identify the most effective solution.

Implementation begins once the project is underway and the scope is defined. At this stage, we follow the project plan while frequently conducting additional research. The first critical step is data exploration, which provides a comprehensive understanding of the dataset. This enables us to effectively prepare and utilize the data for model training.

The next step is feature engineering, which involves analyzing and selecting the key factors or attributes that will guide the development of a model that performs as intended.

Internal work encompasses a variety of tasks, such as conducting academic research and developing internal products.”

What does a day in the life of a data scientist look like?

• 8:00 AM - 9:00 AM: Morning Routine

I typically start the day with a morning routine and having breakfast.

• 9:00 AM - 10:30 AM: Coffee and Daily Tasks Planning

I kick off the working hours by making a cup of coffee and planning out my tasks for the day.

• 10:30 AM - 11:00 AM: Project and Team Stand-Up Meeting

Next, I attend daily catch-up meetings, which may be project-based or with team members. If it's an office day, this time is usually reserved for stand-up meetings.

• 11:00 AM - 7:30 PM: Pre-sales, Implementation, and Internal Tasks

The rest of the day is dedicated to a variety of tasks, including pre-sales, implementation, and internal works, along with meetings with internal teams and clients.

What are the typical tools a data scientist uses?

• Project Management: To track projects and tasks, we use Jira Board. For research or literature reviews, Confluence is used for note-taking and recording analysis results. Also, we record results from data exploration on Confluence for easy sharing with team members.

• Quick Research and Experimentation: In the initial implementation phase, we experiment with concepts, and models, and quickly conduct data exploration using Jupyter Notebook.

• Programming Languages: The primary languages used are Python and SQL.

• Machine Learning Tools: Key tools include PyTorch, TensorFlow, and Scikit-learn.

• Cloud Services: We use Vertex AI and BigQuery.

What do you like the most about being a data scientist?

“First, I love the continuous learning that comes with working in this fast-evolving field. As technology advances rapidly, staying updated is essential. We constantly have to learn because new models are developed all the time. For example, there’s a new version of LLMs almost every month, and we need to understand how they improve and change. Lately, generative AI has been very popular, and cloud providers are continually releasing new tools, so we have to keep up with them which is an exciting part of our job.

Another thing I find rewarding is to see how our knowledge translates into impactful business solutions, while in other scientific fields, the connection to business is often distant. I find that very satisfying.”

How do you see yourself in the next 2 years?

"I see myself continuing as a data scientist, but I want to dive deeper into MLOps to effectively manage the entire data science lifecycle on my own."