AI App Store is the best way to operationalize AI/ML applications built with H2O Wave.
The end goal of data science projects is to produce analytical software applications to facilitate decision-support and automated decision making.
The primary purpose of such applications is to help stakeholders make better decisions by giving them relevant information in an easily understandable format. Most of the heavy lifting is already taken care of by an app’s authors: what data to use, which algorithms to apply, what information to present, and how to present it.
Developing and deploying such applications presents some unique problems:
- Infrastructure. AI/ML modeling is storage and compute intensive. Incorporating machine learning into the software development process and integrating machine learning models into software applications is significantly more complicated compared to conventional software development.
- Talent. Building applications requires a cross-disciplinary team with specialized skills - data scientists, data engineers, backend/frontend engineers and IT/operations - working in close collaboration with stakeholders.
- Time to market. Application requirements are rarely set in stone. Market conditions, competitor offerings, and customer expectations change all the time. Software development teams no longer have months or years to develop and deploy applications. There is an intense need to prototype quickly, gather early feedback from stakeholders, and improve iteratively or fail fast.
In other words, it requires extraordinary effort from a diverse team to wire up data, libraries, tooling and infrastructure before we can focus on what matters most: getting decision-support into the hands of stakeholders.
What is AI App Store?
H2O.ai’s Hybrid Cloud App Store is a turnkey platform that streamlines this entire process: one platform and one API.
- Turnkey Infrastructure. Provides all the building blocks and services necessary to develop and deploy analytical applications in one install. Combines data connectors, data storage, automatic machine learning, model operations and rapid web application development into a single, scalable, vendor-neutral platform with a coherent, end-to-end API.
- Faster time to market. Makes it easy to train models and immediately use them in interactive web applications for rapid prototyping and sharing with end-users. Dramatically simplifies and speeds up the iterative develop-deploy-feedback cycle.