DEV Community

Monika Obrocka
Monika Obrocka

Posted on

Building a High-Performing MLOps Team from Scratch - Lessons Learned and Best Practices πŸ”§πŸš€

In the ever-evolving world of machine learning (ML), having a robust MLOps team is crucial for implementing ML solutions. This post will share how I built an MLOps team from the ground up, without HR support and with just one operations manager's assistance. In only eight months, we grew the team to eight members!

The Journey

When I was promoted to a director role, I had to start recruiting after two months in my new position. My goal was to hire people from Africa, as they best understand the context in the countries we work in. Twitter was my main platform for reaching potential candidates, and I tagged all influencers and groups from the continent. The response was overwhelming, with 80 applications received within the first three days.

The Talent Hunt

While I wanted data engineers, budget constraints led me to seek data analysts with a passion for data and a growth mindset. This mindset indicated their potential to develop the required skills over time. Though candidates might not possess all the necessary skills immediately, their strong desire to grow in that direction was enough for me.

The Screening Process

Screening all candidates was a monumental challenge. I conducted several rounds of evaluations, and during the second round, I identified applicants who would become longstanding members of my team. My interview process began with initiating conversations with promising candidates before presenting them with a technical and theoretical assignment.

The theoretical part involved brainstorming methods for matching two disparate datasets, providing insight into their problem-solving abilities and creativity. The technical assignment was intentionally kept simple, allowing candidates to showcase their expertise in various areas such as analytics, programming, system design, and more.

Lessons Learned

One significant lesson I learned is the importance of communication and treating all applicants with consideration. I regret not having enough time to update all the candidates on their status, and this still haunts me in Glassdoor reviews.


Building an MLOps team from scratch can be a challenging endeavor, but with determination, well-thought-out strategies, and an emphasis on a growth mindset, it is possible to develop a high-performing team.

If you have faced similar challenges while building an MLOps or data team, or if you have used specific strategies and best practices, please share your thoughts in the comments below.

Top comments (0)