An interview with a data scientist

4 min readMar 17, 2023

In this interview, we reached out to one of our customer’s data scientists to talk about their experience working with data, as well as their insights into MLOps and the challenges involved in deploying machine learning models in production. We discussed the daily challenges of dealing with incomplete or messy data, as well as communicating findings to stakeholders who may not have a background in data science. We also delved into the challenges of deploying machine learning models in production, including issues like data drift and model drift, and the engineering challenges involved in designing and implementing a scalable and cost-effective infrastructure to support machine learning models.

Interview with a data scientist

Qwak: Hi there! Can you tell us a bit about your role as a data scientist?

Data Scientist: Sure, I work with large sets of data to extract insights and make recommendations for my organization. This involves a lot of statistical analysis, programming, and data visualization.

Qwak: Great! So what would you say is your biggest challenge on a daily basis?

Data Scientist: I think my biggest challenge is dealing with messy, incomplete data. It’s often difficult to extract meaningful insights when the data is missing values, has errors, or is poorly formatted. This is especially challenging when working with data from multiple sources, as each source may have different data quality issues.

Qwak: How do you deal with this challenge?

Data Scientist: Well, it’s important to start by understanding the data and the context in which it was collected. This can involve working closely with data engineers to understand the data pipeline and identify any potential issues. I also spend a lot of time cleaning and transforming the data to make it usable for analysis.

Qwak: That sounds like a lot of work. Do you have any other challenges?

Data Scientist: Another challenge I face is communicating my findings to stakeholders who may not have a background in data science. It’s important to be able to present complex information in a way that is understandable and actionable for decision-makers. This involves not just creating clear and concise visualizations, but also being able to explain the limitations and uncertainties of the analysis.

Qwak: That’s a great point. Do you have any tips for other data scientists facing similar challenges?

Data Scientist: Sure, I would say it’s important to have a good understanding of the business context in which the data is being used. This can help you identify which insights are most relevant and useful for decision-makers. It’s also important to be comfortable with uncertainty and to be able to communicate that uncertainty effectively. Finally, it’s always helpful to have a good network of colleagues and mentors to bounce ideas off of and get feedback on your work. With data specifically, I find that feature stores provide a good answer to most of the mentioned challenges. It allows you to aggregate, organize and visualize data.

Qwak: Thank you for sharing your insights. Shifting gears a bit, could you tell us about your experience with MLOps and the engineering challenges involved in deploying machine learning models?

Data Scientist: Of course, I have some experience working with MLOps and I can definitely speak to some of the challenges involved.

Qwak: Great. So, what would you say are some of the biggest challenges when it comes to deploying machine learning models in production?

Data Scientist: There are several challenges that come to mind. One of the biggest is ensuring that the model is performing consistently and accurately in the production environment. This means accounting for factors like data drift, model drift, and ensuring that the model is able to handle edge cases and unexpected inputs.

Qwak: That makes sense. Can you tell us more about how you handle these challenges?

Data Scientist: Sure. One approach is to use techniques like A/B testing and canary releases to test the model in production before deploying it fully. It’s also important to have good monitoring and alerting in place so that any issues can be caught quickly and addressed. Additionally, having a solid testing and validation process in place before deployment can help catch any issues early on.

Qwak: That’s helpful. How about challenges with engineering and infrastructure?

Data Scientist: There are definitely challenges involved in setting up the right infrastructure to support machine learning models in production. This can involve things like setting up data pipelines, managing compute resources, and ensuring that the right dependencies are installed. It’s also important to consider issues like scalability and cost when designing the infrastructure.

Qwak: I see. How do you go about addressing these challenges?

Data Scientist: Well, having a good understanding of the underlying infrastructure is important. This means working closely with data engineers and DevOps professionals to design and implement a scalable and cost-effective infrastructure. It’s also important to stay up-to-date with best practices and emerging technologies in the field to ensure that the infrastructure is both reliable and efficient.

Qwak: Thank you very much for taking the time to speak with us and share your valuable insights, experience, and expertise in the field of data science and MLOps. Your perspective on the daily challenges of working with data, as well as the engineering challenges involved in deploying machine learning models in production, has been incredibly informative and helpful. We appreciate your willingness to share your knowledge with our audience, and we look forward to learning more from you in the future. Thank you again for your time and expertise!

About Qwak:

Qwak is a fully managed, accessible, and reliable ML Platform. It allows builders to transform and store data, build, train, and deploy models, and monitor the entire Machine Learning pipeline. Pay-as-you-go pricing makes it easy to scale when needed.




A fully managed AI platform that unifies ML engineering and data operations.