andreas-grafberger/fedless

FedLess is a 2021 research artifact that explored the intersection of serverless computing and federated learning. While the conceptual approach of using ephemeral functions to scale FL training is valid and was novel at the time of publication, the project has failed to gain any significant community traction (14 stars, 2 forks). With zero velocity over the last several years, it remains a static reference implementation rather than a living tool. From a competitive standpoint, it faces extreme pressure from two sides: 1) Established FL frameworks like Flower (flwr.dev) and OpenMined's PySyft, which have vastly larger ecosystems and have matured to support various deployment backends, including serverless-like configurations. 2) Cloud providers (AWS, Google, Azure) who are increasingly integrating federated learning capabilities directly into their ML platforms (e.g., SageMaker, Vertex AI). The 'serverless' moat is thin, as any competent FL framework can be adapted to run on FaaS. Given the lack of updates and small user base, the project is largely obsolete for production use and serves primarily as a citation for its associated paper.