Case study

AI monitoring of KPIs for an online gaming platform



Our client is a large gaming provider specialized in online gaming. They were looking for a service provider who can develop a tool that will help them ensure higher availability and reliability, with real-time response in case of errors or issues that might occur, thus reducing the strain on their operators and improving their response time.

Their main need was to implement a new solution in their business, that would allow quick and efficient responses, without severe time and resource expenditures – resulting in raising overall productivity.

Project goal

Our primary goal was to implement a new solution in our client’s business and develop a tool that would enable the platform to quickly detect anomalies and respond to problems faster, ensuring higher overall reliability and availability of the platform.


The main challenge we faced was having to work with an unlabelled dataset. This was a big obstacle, as it opened a new paradigm in machine learning, and we approached it from an unsupervised learning perspective. It was a novel approach, but we developed the solution successfully.


The developed system is currently deployed internally at Comtrade Gaming, as well as being used by an external client for monitoring the installed online gaming platform.

The entire gaming platform is deployed as a service. The specific anomaly detection and monitoring software aspects of the package have not been deployed as standalone services yet, however, the entire system as a whole is currently deployed and in use by several clients.