Machine Learning Engineer

ABOUT SOFASCORE

Sofascore is a sports-tech company created with one goal in mind – giving sports enthusiasts a deeper understanding of the game.

Our platform is the leading provider of advanced sports insights. From the biggest derbies to amateur matches, every game counts - that’s why we have the largest data coverage with 20,000+ tournaments across 25 sports. This comes easy with Torneo, our very own tournament management software for lower leagues.

The global recognition of the Sofascore Rating, along with the Player of the Season award for the highest-rated players, positioned us as the authority in evaluating player performance.

The Sofascore team counts more than 300 experts in 20 teams, primarily playing at our home court in Croatia, but we also have talents showing their skills worldwide.

More about the company /// More about the platform
ABOUT THE ROLE

At Sofascore , we build products used by tens of millions of people worldwide, turning large volumes of sports and user data into meaningful experiences.

We are looking for a Machine Learning Engineer to join our AI Team and work on real-world ML problems across recommender systems and personalized feeds, semantic search, sentiment analysis, NLP, LLM-powered applications, and other machine learning use cases, potentially including computer vision.

This is an engineering-focused ML role. We are looking for someone who goes beyond experimenting with models in notebooks: someone who can take an ambiguous product or business problem, understand the data behind it, choose an appropriate approach, and turn it into a well-engineered ML solution.

You do not need to have worked on every type of problem we solve. What matters is a strong foundation in machine learning and software engineering, hands-on experience with modern ML and LLM systems, and the ability to think critically about which approach is appropriate for a given problem.

You will work on both new and existing ML systems and collaborate closely with engineers, analysts, and product teams. At Sofascore 's scale, technical decisions have real consequences: scalability, latency, reliability, computational cost, and maintainability all matter alongside model quality.

We prefer candidates who are able to complete the onboarding process from our Zagreb office. Following successful completion of the onboarding process, the role can be performed fully remotely.
Your Responsibilities

- Design, develop, evaluate, and improve machine learning solutions for real product problems used by tens of millions of users worldwide

- Work on a broad range of ML use cases, including recommender systems, personalized feeds, semantic search, sentiment analysis, NLP, LLM-powered applications, classical machine learning problems, and potentially computer vision

- Take ownership of ML problems end-to-end: from understanding the business problem and exploring the data to selecting an approach, building and evaluating models, and collaborating on their integration into our systems

- Develop new ML solutions while also improving and maintaining existing systems

- Build LLM-powered solutions and contribute to the design of retrieval-augmented generation and other modern NLP systems

- Write clean, maintainable, testable, production-quality Python code following sound software engineering practices

- Work with large datasets using Python and SQL, and build reliable data and model workflows

- Evaluate models rigorously, select meaningful metrics, identify issues such as overfitting and data leakage, and understand the trade-offs behind different modelling approaches

- Collaborate with product managers, analysts, software engineers, and other stakeholders to translate product and business needs into well-defined ML problems

- Critically evaluate proposed solutions and choose the right level of complexity for the problem, whether that means a simple heuristic, classical ML, deep learning, a recommender system, or an LLM-based