Federated Learning
Also written as Federated ML, Privacy-Preserving ML
Training a model across many devices or organisations without moving their data to one place. Each device learns from its own data and sends back only model updates. Used for phone keyboards, and for hospitals or banks that cannot legally share raw records.
Think of it like
Several chefs each improving a shared recipe in their own kitchens and posting back only their notes, never their ingredients.
Junior or senior?
Rare, specialised experience, usually from big tech, healthcare or finance research teams.
Senior sounds like
Talks about devices with very different data, unreliable connections, and how they checked the updates themselves didn't leak private information.
Ask them
“Where was the data sitting, and why couldn't it be moved? What made training harder than doing it centrally?”