Design an Online Experimentation (A/B Testing) Assignment Service

You are asked to design a highly available online experiment assignment service. The service exposes an API that takes (userId, featureName) and returns a variant identifier (e.g., control or treatment). Assume the service will be used by web and mobile applications globally and must support both logged-in and logged-out users.

Clarify, specify, and propose designs for the following:

1. Deterministic sticky assignment and bucketing strategy
2. Configuration storage for experiments and features
3. Eligibility targeting and audience filters
4. Traffic allocation and ramps
5. Mutual exclusion and experiment layering
6. Overrides (holdouts, QA, killswitch)
7. Exposure logging and idempotency
8. Metric tagging and data pipeline
9. Privacy and compliance (e.g., GDPR/CCPA)
10. Latency/availability targets and fallbacks
11. Experiment lifecycle (create, start, pause, stop, analyze)
12. High-level architecture and key components

Provide a high-level architecture, key APIs, and justify key trade-offs. Where relevant, include small examples and precise definitions (e.g., hashing and bucket math, namespace/layering).