As the Internet becomes more prevalent in our daily lives through new devices and services such as IoT and Cloud, cyber-attacks have become more sophisticated. Traditional approaches to cyber security such as perimeter protection and signature or rule based detection systems are no longer effective in detecting and preventing such attacks. Using large amounts of data collected in the form of logs and statistical modelling techniques such as Machine Learning, it is possible to develop highly accurate androbust threat detection systems. In this talk I will give an overview of how Data Science and Machine Learning are used to build cyber threat detection systems and present some examples from the Industry.