How to Build High-Performing Machine Learning Teams?

Posted on 16 June 2023

​Machine learning has become an indispensable tool for businesses looking to gain a competitive edge in today's constantly evolving data-driven world. However, building a high-performing machine learning team is not an easy task. It requires a combination of technical expertise, creative problem-solving skills, and effective teamwork.

In this article, we'll explore the key factors that contribute to the success of a machine learning team. From what essential skills are needed, to fostering the right environment, we’ve spoken to half a dozen industry experts and leaders in this space who have kindly contributed their thoughts.

Whether you're an experienced manager within machine learning or just starting out on your journey within management, this article will provide you with valuable insights on how to create a high-performing machine learning team. Let's get started!

Firstly, a thanks goes out to those who were kind enough to spare some time to give us their thoughts:

  • Adrian Matei – Head of Data & ML at GrowUp

  • Sumanas Sarma – ML Engineering Team Lead at THG

  • Rishabh Mehrotra – Director, Machine Learning at Sharechat

  • Javier Rodriguez Zaurin – Director of Machine Learning at Assetario

We were also joined by a Machine Learning Product Lead, who has asked to remain anonymous.

 

Our first consideration is what skills are essential for any high performing machine learning team?

We asked Adrian Matei for his thoughts on this and he mentioned that before starting any machine learning you need to make sure your data is stored in a suitable environment such as a data warehouse or a data lake. Ideally, you’d also have a machine learning platform for ML pipelines. Once the correct storage and pipelines are in place, that’s when you would need your Data Scientist, who can build domain-specific models and algorithms to tackle specific business critical challenges.

“The team need to be aligned to the rest of the company. Aligning with the specific business needs and ensuring they build solutions around these needs, rather than the other way around. Once you’ve understood the business’s needs, it’s critical you design specific success metrics and work towards these.” – Machine Learning Product Lead.

Javier Rodriguez Zaurin and Sumanas Sarma both agree that having a team with a diverse background will enable you to solve more complex problems and have a great blend between scientific rigour, a curious mindset, and coding efficiency should go a long way to achieving strong results.

The second question is centred on what characteristics do you look for when hiring for your team?

All our participants mentioned that the one of the main things they look for when hiring into their team, is whether that individual has delivered real value to their previous companies and can they concisely explain the input and value they delivered on projects.

Rishabh Mehrotra mentioned that when he typically hires for his team that he looks for individuals that have deep expertise in relevant problem spaces and therefore have opinions on subjects and can bring fresh ideas and can drive that charter.

Sumanas looks for highly motivated and enthusiastic learners as they will be more useful than an experienced coder who isn’t perhaps as motivated but mentioned that is only if you can provide them with the right resources to train them. Also, he goes on to add that the ability to build rapid prototypes, scoping work out early and often, will help keep business stakeholders engaged and ensure you don't alienate them from what is often seen as complex topics.

The third question we asked the ML leaders was How do you ensure that your team produce excellent outputs?

Adrian outlined that the ML products you and your team are building should be solving a business problem. Also, at the start of the process, you need to understand the requirements of the business and get the stakeholders brought in quickly. Finally, the team should have clear metrics for success and work in sprints to achieve these goals.

Rishabh was of the opinion that a number of different factors can contribute to excellent output, such as a clear-cut roadmap, outlining the baseline, iteratively working through the roadmap, predictability of the system, system robustness and finally, similarly to Adrian, being clear on what a successful metric looks like.

The final question we asked our panel is what advice you’d give hiring managers starting out on this journey?

Javier advised that at the start of your tenure with an organisation, you should sit down with the leadership team and quickly align on business requirements and collective goals. In addition to this, he suggests that getting to know your peers and fellow leadership team quickly will enable you to build relationships across teams.

“New hiring managers should be aware of the timescales it takes to hire, but then also how quickly it will take any new starters in the ML team to get up to speed, and then work your hiring plans backwards from there. I believe that you should align individual goals with the overall team goals, so that individuals are motivated to work towards a common goal. Finally, as a hiring manager, you should aim to delegate as much work as possible to the relevant individuals in the team.” – Machine Learning Product Lead.

Conclusion

I hope you enjoyed reading the Q&A! I’d like to thank Adrian, Sumanas, Rishabh, Javier and the ML Product Lead for contributing to the article and allowing me to publish it to my network. Please check out their LinkedIn profiles to see what they are up to and connect with them directly for more nuggets of useful information:

https://www.linkedin.com/in/adrian-matei-a8218661/

https://www.linkedin.com/in/sumanas-sarma/

https://www.linkedin.com/in/mehrotrarishabh/

https://www.linkedin.com/in/javier-rodriguez-zaurin-06277454/

Share this article