On Device Privacy Architecture Questions

Describe designing systems that minimize data collection and prioritize local processing on devices when appropriate. Topics include privacy preserving techniques such as federated learning and differential privacy, secure hardware enclaves and key management, local encryption and secure update mechanisms, data minimization and retention policies, and trade offs between functionality and privacy or regulatory requirements. Discuss engineering challenges of on device model inference including model size reduction, quantization and pruning for resource constrained devices, secure telemetry and diagnostics, and methods to measure privacy and utility trade offs.