Flower is fundamentally a collaborative effort led by our open-source community. Contributions take many forms, from the feedback regarding Flower features to full algorithm implementations like those we received during our Flower Summer of Reproducibility initiative.
Pilot Program
But alongside these organic open-source development efforts, we have also found establishing more structured partnerships with individuals and organizations helpful. We call these collaborations pilots and run them as part of the Flower Pilot Program. Such pilots take many forms. In particular, they help tackle multi-stage projects that can take a while to produce results, such as working together to build an open-source prototype of a new form of federated AI system and or developing a brand new feature in Flower, especially when it is unclear how to balance simplicity and power in the API design. Importantly, through these interactions we find we learn a lot about how users approach federated learning and how Flower can help them best accelerate their work.
Success Stories
The Flower Pilot Program has operated for over a year. Already, we have several success stories. We have worked with organizations attempting to scale their Flower deployment to over 100M smartphones. Others have transitioned from simulations built only for a research paper to a Flower-based deployment that can process patient data in a health setting. We have also worked with organizations like Brave towards hardening support for generally valuable capabilities, such as efficiently running a Flower node within a Chromium browser. Most critical, however, has been the collective insights we learned with our Flower users that were fed back into the design of Flower Next, recently launched at Flower AI Summit 2024.
Batch Two
Today, we are launching the call for the next wave of participants in the Flower Pilot Program. Full details of the program process are available from our pilot website. We have expanded our capacity to handle pilot collaborations. However, we will still focus on engaging with a few genuinely ambitious partners who want to do things in federated learning that are considered impossible or accelerate the deployment of federated systems that can disrupt the world. Our mission is to have federated learning become the de facto standard approach in machine learning, and we would like to work with people and organizations that share this vision.
We look forward to reviewing pilot applications for this second batch and working with you. You can begin the application process by following this link.