Lithops is a map/reduce engine that executes on top of serverless platforms. Both simple to use and platform independent (Python library) and cost-effective (because of serverless) –only pays for actual computation cost, not cluster cost.
Main benefit: focus on customers’ data science algorithm and use Lithops to launch customers’ computation at scale in the cloud. The plumbing (underlying platform) can be delegated away from the data scientist.
Addressing target audiences and expressing needs
- Collaboration
Collaboration and synergies with other projects and organizations
- Research and Technology Organisations
R&D, Technology and Innovation aspects
the Lithops Novelty is being managed by IBM outside CLASS, in other projects. Recall that Lithops is a long-time open-source (based originally on Berkeley’s PyWren project), so it cannot be used to generate new Intellectual Property (IP).
Lithops is already being used by an IBM Cloud customer with some enhancements contributed from CLASS. Going forward, IBM studies the use-case and expands to additional customers.

