From Data Chaos to Actionable Insights: My Team’s Quest to Hire our First Data Scientist

Editor’s Note: This article was guest-authored by CoderPad’s Senior Principal Engineer Jonathan Geggatt about his data science hiring experience with his previous company, HotelTonight.

In the fast-paced business landscape, the ability to harness data effectively can be a game-changer. 

This became quickly apparent to me during my time as the Director of Data Science and Engineering at HotelTonight, when my team and I identified specific challenges that needed the expertise of a data scientist. Our goal was simple: to transition from a reactionary data team to one that could proactively generate actionable insights and keep a finger on the business’s pulse.

Identifying the Need for a Data Scientist

Initially, our data team functioned in a reactionary mode, scrambling to respond to requests for data insights. This approach often resulted in missed opportunities for strategic planning and proactive decision-making. To counter this, we embarked on a journey to build a data science team that could unravel ambiguous tasks and extract valuable insights to steer the business in the right direction.

🔖 Related resource: Jupyter Notebook for realistic data science interviews

Crafting a Comprehensive Hiring Process

The first step in our journey was to establish a hiring process that could identify candidates who aligned not only with the technical requirements but also with our organizational culture. Our Talent Acquisition team took charge of the initial screening, filtering resumes based on interests and salary expectations, bypassing the technical pre-screen to dive deeper into potential fits.

There would then be a phone screen where the hiring manager introduced the candidates to the team dynamics and the expectations tied to the role. Following this, we conducted a technical interview to assess the foundational skills in SQL and Python. A cultural fit interview ensued, allowing us to gauge how well the candidates could integrate into our organizational ethos.

The highlight of the process was the data science project exercise, where we presented candidates with high-level questions crafted by our team. This stage was crucial in evaluating a candidate’s ability to handle real-world scenarios using tools like Snowflake and Jupyter Notebooks, offering a realistic and immersive experience with real, anonymized, and scrubbed data.

We would spend about 20 minutes going over the question with them, including an overview of the data dictionary. We also gave them the code for a database connection to reduce the amount of time they spent on trivial tasks.

They would then spend one-to-two hours on the task, and present the results to us as screenshots or in a Jupyter Notebook.

🔖 Related read: Mastering Jupyter Notebooks: Best Practices for Data Science

Learning and Adjusting Along the Way

In the early stages, we faced the challenge of not knowing the “right” answers to our high-level questions — a gap we were hoping to fill with the new hire. The criteria we set for evaluating the candidates were centered around their past projects and their ability to convey complex insights to non-tech stakeholders. We were cautious about candidates who ventured into building predictive models, considering their limited knowledge about our business.

The Evolution of the Hiring Process

As the data science team took shape, our hiring process matured. We could now focus on the techniques used by candidates to tackle problems, expanding our hiring spectrum to include fresh graduates and individuals transitioning into data science from unique backgrounds. This transition also equipped us with the discernment to sift through the applicants effectively, identifying those who could genuinely add value to our business as opposed to those who merely echoed what clients wanted to hear.

Leveraging Technology for a Competitive Edge

The integration of Jupyter Notebooks in our hiring process turned out to be a significant advantage. By allowing candidates to use familiar tools, we set them up for success, giving them the opportunity to showcase their best work. The use of our Snowflake data set not only offered a realistic experience but also allowed us to flaunt our tech stack, giving us an edge in attracting top talent.

Key Takeaways: Finding the Right Blend of Technical and Soft Skills

As we navigated this journey, it became evident that crafting questions aligning with our business objectives was crucial in finding the right person. The ultimate role of a data scientist in our team extended beyond technical expertise to include the ability to simplify complex data for stakeholders, making personal and communication skills a vital aspect of the selection criteria.

In conclusion, my journey at HotelTonight in hiring our first data scientist was one of learning and evolution. As we move forward, the focus remains on finding individuals who can decomplexify data, turning it into actionable insights and fostering a proactive, data-driven business culture.

Improve your own data science interviews

CoderPad makes it easy to assess data science candidates with our new Jupyter Notebook-integrated pads. Check it out for yourself in the pad below.