Chapter 1
Local vs. Cloud AI models have significant differences.
1) Scope & Deployment
Local models: Run on devices or on-premises servers under direct control of the user or organization. Deployment includes laptops, mobile devices, edge gateways, or private data centers.
Cloud models: Hosted and executed on remote provider infrastructure. Accessed via APIs, managed services, or hosted platforms.
2) Performance
Local: Lower latency because inference happens on-device or on-prem; suitable for real‑time interaction, robotics, augmented reality, and IoT.
Cloud: Higher latency depending on network conditions; acceptable for batch jobs or interactive use when network latency is small relative to processing time.
Private vs. Public Clouds
3) Privacy and Data
Local: Stronger data sovereignty and privacy because data can remain on-device or within private infrastructure. Easier to comply with strict regulatory or internal policies.
Cloud: Data sent to provider for processing; requires careful contractual, encryption, and governance measures to meet privacy and compliance requirements.
4) Security
Local: Reduces attack surface associated with network transmission; however, device security and physical access controls are critical. Patching and model integrity must be managed locally.
Cloud: Providers offer advanced, centralized security controls, monitoring, and managed patching, but risks include misconfiguration, multi-tenant exposure, and third‑party access.
(Optional)More Information…
-
Our downloads have everything you need to supplement this course.