Ryax Documentation

Welcome to this Ryax documentation!

This is the best place to start with Ryax. We cover what Ryax is, what problems it can solve, and how it compares to existing software. If you want to get started as fast as possible, writing modules and workflows jump directly to the crash course.

What is Ryax?

_images/ryax-main.jpg

The data revolution is happening now. Outstanding results are achieved using data to predict the future, automate tasks, and visualize the impossible. Indeed, all major players exploit huge amounts of data applying complex algorithms in an industrialized way. Data is flowing from everywhere and storage is affordable. As a consequence, we want to save all data available in our range: social media, IoT devices, sensors, you name it. Considering that, applying schemas to data is becoming hard or sometimes unfeasible: takes time, associated value to data is unknown. Data storage schemes are shifting from data warehouse (schema-on-write approach) to data lakes (schema-on-read). In a nutshell, we store data as is, without minimum formatting and cross-referencing.

Data scientists create algorithms that crunch data to harness business’ value. Many libraries for common inference like TensorFlow, PyTorch, to cite a few, help in this context. However, there is a lack of standards to define applications in this constantly evolving scenario. Solutions are often ad hoc, and industrialization is done on a case-by-case basis. But algorithms have many aspects in common. For instance, grab data from a source API, apply a data frame schema, analyze the data frame, and produce visualizations. These algorithms should fit a common standard to be reused, automated, and integrated. Like that a complex infrastructure can adapt behind the scene to the load needs seamlessly given data scientists the best user experience.

Ryax is the answer to your data science industrialization problems. No more ad-hoc solutions that only a single person understands. Finish the problems of deploying and running periodically. Easy reuse of pieces of software once converted to Ryax modules. Support the most powerful tools for data scientists: pytorch, TensorFlow, matplotlib, and more. Ryax is a platform for Data Teams (Data Scientists, Engineers…), helping them transform, automate, and deploy their data projects into business opportunities.

Some of Ryax’s key benefits:

  • Automation: auto deployment, multi-node orchestration, native error management and resume, data consistency management… Ryax robustly automates your Data Science execution.

  • Observability: monitor any point, report effortlessly internally and externally with Ryax’s integrated features (execution logs, multi-node monitoring and full-traceability)

  • Evolutivity: naturally grow towards new analytics (ML, AI…), new infrastructures (Hybrid Cloud, Edge-to-Cloud) and best practices (CI/CD processes, code modularity…)

  • Low code: drop your models, scripts and algorithms in the Ryax platform, we handle the rest: building, packaging, deploying, updating…

  • Event based: Ryax’s workflows react instantaneously to any events and even combine events with streaming operators

  • Built for heterogeneous infrastructures: easy processing placement across complex IT infrastructures, transparent data transfers between modules…