Question 1 of 25

Model serving in a GenAI product usually means:

Question 2 of 25

Why do many teams start with a hosted LLM API before self-hosting?

Question 3 of 25

Why is caching important in LLM apps?

Question 4 of 25

A good Redis cache key for a RAG answer might include:

Question 5 of 25

Rate limiting an LLM API route primarily protects against:

Question 6 of 25

A token bucket rate limiter allows:

Question 7 of 25

How can you reduce token cost in production?

Question 8 of 25

Prompt caching (provider-side) helps when:

Question 9 of 25

What is often the latency bottleneck in GenAI apps?

Question 10 of 25

Streaming responses to the UI mainly improves:

Question 11 of 25

Production monitoring for LLM apps should track:

Question 12 of 25

Structured logging for each chat request should include:

Question 13 of 25

When an upstream LLM API times out, good error handling:

Question 14 of 25

Graceful degradation might mean:

Question 15 of 25

The Module 10 capstone combines RAG + agents + UI with production concerns. The main goal is:

Question 16 of 25

LLM API keys should live:

Question 17 of 25

You re-index blog posts for RAG. Cache keys must:

Question 18 of 25

HTTP 429 Too Many Requests tells the client:

Question 19 of 25

Output tokens are often more expensive per token because:

Question 20 of 25

Model routing in production might:

Question 21 of 25

Time to first token (TTFT) measures:

Question 22 of 25

A sudden drop in cache hit ratio after deploy might indicate:

Question 23 of 25

Exponential backoff on retries means:

Question 24 of 25

Streaming LLM responses to the browser requires:

Question 25 of 25

Including requestId in logs and user-facing errors helps: