Machine learning

When you’re trying to extract information from large amounts of complex, real-world data, chances are you’ll need to use machine learning.

We are currently using machine learning with really diverse datasets, which are big and messy. Our customised models and methods are helping clients find themes in large amounts of text, analyse forest satellite data and work on speech recognition.

While more data means we can use bigger models and get better results, large models quickly become an engineering problem. There’s a challenge for us here to write code that interacts with the specialised hardware very efficiently.