Neuro

The product is a serverless API for ML compute that lets users train, run predictions, and deploy machine learning models without managing infrastructure, handling GPU provisioning, scaling, and bandwidth automatically. It is for developers and data teams in businesses that build and operate ML models in Python environments such as Jupyter notebooks. It is delivered as a pay-as-you-go API and Python library that connects to top-tier GPU servers and stores data in AWS S3, requiring only a few lines of code to use.

Key features

• Train ML models via API
• Run predictions via API
• Deploy models to production
• Auto-assign fastest GPU per task
• Support PyTorch TensorFlow MXNet
• Scale infrastructure automatically
• Pay only for compute time
• Store datasets and models on AWS S3
• Track metrics with automatic visualisations
• Parallel model training
• JSON support for predictions
• Run from any Python environment