Zenaique
Topics
Practice
Browse
Reference
Challenges
Pricing
Search…
⌘K
Topics
/
LLM System Design
/
Ops
Ops
Subtopic
10 questions
Questions tagged with Ops — part of LLM System Design.
Format
Difficulty
Role
Experience
Companies
Sort
Newest
Quality
Difficulty ↑
Difficulty ↓
Questions
What belongs in the trace of every LLM call?
Multi-select
Medium
·
Qual 4.0
How should a high-throughput service absorb provider 429 responses?
Short Answer
Medium
·
Qual 4.0
Why production prompts get versioned and model IDs get pinned
Flashcard
Easy
·
Qual 4.0
Which guardrails belong on the output path of a customer-facing endpoint?
Multi-select
Medium
·
Qual 4.0
Design the isolation a multi-tenant LLM platform needs between customers
Short Answer
Hard
·
Qual 4.0
What does an LLM gateway in front of providers centralize?
Short Answer
Medium
·
Qual 4.0
Why gate every prompt or model change behind a held-out eval in CI?
Short Answer
Medium
·
Qual 4.0
Why capacity-plan an LLM service in tokens per second, not requests per second
Short Answer
Medium
·
Qual 4.0
A GPU pool sustains 9,000 output tokens/sec — will it hold this traffic?
Predict Output
Hard
·
Qual 4.0
A downstream API changes its schema while your agent is live in production. How do you handle it?
Short Answer
Hard
·
Qual 4.0
OpenAI