ONE FOR THE PRODUCT MANAGER:
ENGINEERING vs DATA SCIENCE
Machine learning (ML) is a powerful methodology that can help any corporation boost their decision making capabilities. Any piece of software can use machine learning to enhance your team’s output and help them accomplish more with less effort, reducing operational costs, increasing revenue and even modernising the relevant industry.
The success of this kind of apps not only relies on well-designed models but also on a well-designed backend, user experience as well as the interconnections between all of these components (DS-BE-UI). This interaction is what we call the:
ML-PRODUCT CHALLENGE: ENGINEERING VS DATA SCIENCE
For any usual software product, a backend and a user interface need to be developed. This implies the Engineering team working closely to the UI and user-experience designers. As a Product Manager, you have to know that any ML product adds one more level of complexity: the modelling. And with this, the incorporation of data scientists to your team.
This two disciplines, even though they look similar, they are far away from each other. Engineers works towards stability and repeatability in their processes. However, data scientists work in a research environment, dealing with uncertainty and lack of the latter. It is key to know how to set them both for success in order to succeed in any Data Driven Disruption project.
Here we detail some points any Product Manager should take into account when leading a ML Product team. Engineering vs Data Science:
Engineers in your team are the ones in charge of the stability of the final product. Their needs and environment where they are comfortable derives from that. Used to dealing with changes mainly coming from the designers, now as a product manager, you have to be ready to deal with them receiving inputs also from data scientists.
A good way to get ready for this interaction is to make them understand the skillset of the data scientists you are bringing to the team so their expectations are in line with reality. Lack of testing, code quality under engineering standards and complex mathematical outputs tricky to digest for engineers are just some of the things your team needs to have in mind.
DATA SCIENTISTS: #CHANGES
Usually very smart and fast researchers.
Usually not structured and working in silos.
As a product manager you need to know, data scientists will be delivering two very different outputs:
- Metrics on model performance to a business audience
- Models and code to engineers to be implemented in the product
Business audience needs to be aware that, even mathematical results are delivered, this does not mean they have been embedded in the product at all. They need to know, their validation is required for this to even begin.
For a fast implementation of data scientists’ model, engineers will have to prepare a framework and rules for the data scientists to deliver results in a sustainable way. And always remember:
DATA SCIENCE & ENGINEERING ARE THE SAME TEAM, MAKE IT FEEL THAT WAY