A highly flexible & start-up proof collaboration

After working at Faktion Willem Romanus & Sibren de Preter, both passionate sports and data engineers spun off to build their own start-up venture. Their new quest wasn’t a goodbye, as the two young entrepreneurs sought out the help of our ML engineers to build a state-of-the-art training application. We asked them about the collaboration and their experience as a start-up in working with us on this testimonial.

How would you define the scope of this collaboration?

We wanted to build a different solution from other wearables in the market regarding precision, prediction, and safety for exercise and therapy treatments. Having a junior technical team, we wanted to learn as much as possible quickly while also developing a quality prototype. We asked Faktion to coach, guide, and consult us through that process and took senior ML engineers Jos Polfliet & Jeroen Boeye on board to mentor us.

The project kick-off was already quite productive. Despite a lot of information from our side, Jeroen and Jos could quickly follow and give relevant expertise. We sent an extensive prior briefing by mail, which was thoroughly read before the meeting by Jeroen. This was highly appreciated and made the first meeting already productive.

How do you assess the content of the project?

Faktion took on an advisory role where they brainstormed with the technical team, introduced us to the right algorithms for the task, and coached us on ML product development’s best practices. On a more technical note, we had two layers of classifiers. The first model would detect the exercise. The second was a multi-label classifier that could detect errors in exercise execution (+ combinations of errors). Working out these two classifiers was crucial in our prototype development, and having an experienced team alongside us helped speed up the development process.

Faktion was very proactive in setting everything up so we could work effectively and efficiently. Senior ML engineers from Faktion also reviewed the code. Considering the highly sensitive nature of predictions, they also executed an exploratory analysis of the raw data. Our technical team could then use these analyses to develop new functionalities and features.

How did you experience the collaboration?

We experienced collaboration with Faktion to be highly flexible and start-up proof. Starting from the sales team’s attitude to the responsiveness of the senior consultants to our needs. To us, as a start-up, the flexibility stood out and made the difference. We experienced a customer-first mindset and felt a deep understanding of being in a start-up/uncertain position!

Faktion definitely beat our expectations, even though they were very high from the beginning. We got stunning machine learning expertise & deeply passionate engineers who made a real mission of helping us out. We could reach out & come over at any time and felt “welcome” & heard at all times!

What was the outcome of the project?

We saw good early results in the capability of the models to detect training exercises. Precisely the real-time correctness-analysis is what differentiates this solution from other fitness solutions on the market.

Our model doesn’t only improve revalidation, but it also lays the ground for better individual preventive training and real-time monitoring.

Unfortunately, once we started collecting training data on a large scale when COVID hit, the project was put on hold. Nonetheless, we managed to build a solution that guides, coaches, and trains the users to do the right exercises the right way.

Would you recommend Faktion?

Yes, 100%, and already did many times. ☺