Past Projects

Describe a business problem that you helped solve using data science.

Evaluate ability to deliver when requirements are not clearly defined, ability to work with different functions e.g. product, engineering.

In hindsight, how might you have tackled the problem differently

Look for evidence that the candidate is reflective about their past work and is constantly thinking about how to improve things.

What was the success metric

Do they understand how to formulate business problem into data science problem?

How do you implement this

Understand how the candidate architected the solution from an ML engineering perspective. What frameworks did they use, what were the design considerations, how is failure managed?

How did you plan/manage this project

Does the candidate have the professional maturity to plan & negotiate timeline with product/biz?

Did you face any difficulties, technical or stakeholders

Candidate is able to articulate difficulties they faced and illustrate their thought process how to solve it retrospectively

If you had to redo it, what would you have done differently

ML Engineering

How do you usually manage your environment and project dependencies?

Demonstrates experience with approaches/technologies to ensure reproducibility and understands their importance for DS projects (e.g. virtual environments, docker, makefiles)

Suppose your dataset is too large to fit into memory on a single machine. How would you start to build a feature engineering pipeline for feature engineering and model training in this case?

Is familiar with distributed data processing technologies. Recognises the importance of building a prototype before adding greater complexity.

When would you use SQL for feature engineering and when would you use imperative code?

Understands the distinction between imperative and declarative code and some pros and cons (readability, testability, efficiency, reusability)

How do you approach testing in your data pipelines?

Familiar with various testing paradigms and the specific challenges with testing data pipelines e.g. unit testing, frameworks for automating data quality checks, types of data checks (completeness, uniqueness, freshness etc.)

What do you see as the main challenge in deploying ML models compared to other software applications?

Familiar with production systems to fetch features, ensure consistency with training data, frequently refresh models, manage experimentation, etc.

Business Sense and Data Driven

How do you identify the most relevant stakeholders for a given data analysis project, and how do you ensure your analysis addresses their concerns or needs? How do you tie your insights to stakeholders’ concern?

Give candidate a scenario and ask them to present to stakeholder (interviewer). End results should contain actionable recommendations.

How would you handle a scenario where different stakeholders have conflicting interpretations of your analysis results? i.e. improvement in ETA but drop in BCR.

In what ways do you validate the quality and relevance of the data you use for analysis, especially when it comes from multiple sources?

• Does candidate do some sanity check on data before using it?
• Will they check with the relevant persons who own the data or do they just make assumptions of the data?

If you were asked to prioritize multiple data analysis projects with limited resources, how would you decide which projects to tackle first?

How do you ensure that the methodology you have chosen is accurate? Does this candidate talk about how they find information to support the methodology they’re using or test out their hypothesis?

Bar Raiser

What motivated you to apply as a data scientist to join the company?

[EXPECTATION] Understand the candidate’s career goals and aspirations. To understand what is the candidate’s expectation as a DS.

What are your strengths and weaknesses?

[REFLECTION/AWARENESS] Self-awareness and reflection. Self perception.

Share a time when you had a disagreement with your colleague or manager and how do you handle it?

[CONFLICT] Understand candidate’s conflict resolution skill and how they handle difficult situations.

Describe a time when you had to work with someone with a different working style than yours. How did you handle it?

[CONFLICT] Adaptability and handle working with people who have different working styles.

What are your long term career goals?

[GROWTH/ASPIRATIONS] Understand the candidate’s research skills and interests in the company.

Suppose there are many initiatives within this sprint that you might not be able to pick up and the product manager says everything is important. How do you handle this?

[EXPECTATIONS MANAGEMENT] Understands how the candidate prioritize workload and communicate and manage expectations of stakeholders.

How do you stay up to date with the skill set in data analysis, statistics as well as data science?

[LEARNING] Assess candidate’s motivation for self-improvement.

Share a time where you had to work with a topic that you are not very familiar with and how do you handle it?

[LEARNING] Assess a candidate’s ability to handle uncertainty and strive for self-learning and reference to literature to help solve problems. Understand that no one knows everything but able to learn fast to get things done.

Is there an example where you had to take initiative on a project? Why and how do you handle it?

[OWNERSHIP/LEADERSHIP] Understand candidate’s leadership skills and ability to take ownership in their work.