Our Computational Tools
Speed, Standarization and Stackability
Standardizing our work environments with Docker guarantees repeatability and deployment speed, enabling us to easily decouple processes and stack them as needed.
Computing Scalability and Parallelization (GPU)
Google Cloud's variable computing capacity keeps costs down, while using GPUs enables parallelized computations that speed up our model training.
Flexible, Replicable and Scalable Workflow
Our set of pre-built Jupyter notebooks provides us a common workflow for every project, from data ingestion, to modeling to testing, while still being adaptable to every client’s reality or needs.
State-of-Art Machine Learning Tools
PyTorch provides efficient GPU usage, which significantly speeds up computation for our Bayesian algorithms. This allows us to train even our large-scale models quickly.