Modern deep learning frameworks drastically reduce the engineering effort necessary to train and execute neural network models at scale. We work with a variety of technologies: TensorFlow, Theano, Pytorch and Dynet.
Python provides a sea of useful tools and libraries. We take advantage of the Python ecosystem in most of our machine learning and data science related work.
NodeJS is the technology of choice for web application backend systems.
The core of our large-scale crawling system is written in Java, which strikes a good balance between development effectiveness and performance.
We store and process petabytes of data. The distributed database system Apache Cassandra helps us to ensure linear scalability.
Most of our infrastructure is powered by the Google Cloud Platform and container technology.