Whenever you're ready, walk me through how you'd approach this.
Describe a time you went through system design and implementation deep dives, including an LRU cache, and how you handled the process.
Explain how you clarified requirements, decomposed the design, validated implementation choices, and handled tradeoffs. Be prepared to discuss how the same process applies to an end-to-end ML system, including serving, latency, monitoring, and failure recovery.