❓ Quiz — intermediate/session-08

Question 1 — What is the essential difference between a notebook prototype and a production service?

HAS The prototype uses a less powerful model than the production B Production must manage what the prototype ignores: authentication, failures, latency, costs, observability, security and scalability for thousands of users C The prototype is written in Python, the production in another language D There is no difference if the prompt is well written

Question 2 — Why should you NEVER place an API key in client-side code (browser, mobile application)?

HAS Because it slows down the page loading B Because providers prohibit more than two keys per account C Because anyone inspecting the client's code (developer tools, decompilation) can read the key, steal it and consume the API at your expense D Because API keys only work on Linux servers

Question 3 — Your application receives an HTTP 429 (Too Many Requests) response. What is the correct reaction?

HAS Try again immediately, in a loop, until it passes B Wait then try again with an exponential backoff (1 s, 2 s, 4 s...) and jitter (random variation), respecting the header retry-after if provided C Permanently abandon the request: a 429 error means the account is banned D Change API key to bypass throttling

Question 4 — Which category of errors should NOT be automatically retried?

HAS 429 errors (rate limit exceeded) B 500/503 errors (server error or temporary unavailability) C 400/401/403 errors (invalid request, missing authentication, access prohibited) — retrying the same invalid request will produce the same error D Network wait times (timeout)

Question 5 — Why does streaming via Server-Sent Events (SSE) improve user experience?

HAS It reduces the number of tokens generated, therefore the total latency B It makes model generation faster on the provider side C It compresses the response to save bandwidth D It displays tokens over generation: the total latency hardly changes, but the perceived latency collapses because the first word appears in less than a second

Question 6 — How does prompt caching allow savings of up to ~90% ⚠ on the hidden part?

HAS The supplier charges a very reduced rate for tokens at the start of the prompt that are identical from one request to another (system prompt, documents, tools) — hence the rule: stable content at the beginning, variable content at the end B The cache stores the complete responses, the model is never called again C Caching compresses tokens so that they count for half as much D Caching only saves time, never money

Question 7 — What is model routing?

HAS Distribute network traffic across multiple data centers B Direct each request to the cheapest model capable of processing it: a small model for simple tasks (often the majority of traffic), a large model for difficult tasks C Change LLM provider each month depending on prices D Always use the most recent model

Question 8 — Why do we monitor latency in percentiles (p50, p95, p99) rather than on average?

HAS Percentiles are simpler to calculate than the average B The average is always higher than the p99 C The average hides the tail of the distribution: a service can have a good average and a catastrophic p99 — but it is the users in the worst cases who complain. SLAs (Service Level Agreement) are written in percentiles D Monitoring tools do not know how to calculate an average

Question 9 — What does “prompts are code” mean in practice?

HAS Prompts must be written in a programming language B The prompts are versioned (Git), reread by peers and tested: with each PR (Pull Request, merge request) which modifies a prompt, a series of evaluations runs in CI/CD (Continuous Integration / Continuous Deployment) to detect regressions C Prompts must be compiled before being sent to the model D Only developers have the right to modify prompts

Question 10 — You must summarize 50,000 documents every night, without a human in front of the screen, with a supplier rate limit of 500 requests/minute ⚠. Which architecture to choose?

HAS Send all 50,000 requests at once and retry those that fail B A dedicated GPU (Graphics Processing Unit) instance, essential for this volume C A queue architecture: documents are placed in the queue, workers consume them at the rate limit, the results are stored for the morning D SSE streaming, so that summaries arrive faster

Question 11 — What is quantization of a model, and what is it for?

HAS Limit the number of requests per minute to stay under the rate limit B Reduce the number of bits by weight (e.g. FP16 → INT8 → INT4): the model occupies less memory and reads faster, at the cost of a slight loss of quality — it’s “the JPEG of models” C Compress prompt tokens so that they cost less to the API D Training a small model to become as good as a big one

Question 12 — You need to self-host an open-weights model of 70 billion parameters (data sovereignty) on two 24 GB VRAM cards. In FP16 it weighs ≈ 140 GB. Which approach is the most reasonable?

HAS Impossible: you must go through a supplier’s API B Load it as is in FP16: the 140 GB will fit on 48 GB of VRAM C Quantify it in INT4 (Q4, format GGUF) : it drops to ~40 GB, fits on both cards, with generally minimal quality loss (1–3 points ⚠) D Cut it into 70 small models of a billion parameters executed one after the other

Question 13 — In an enterprise AI architecture, what does Zero Data Retention (ZDR) actually mean?

HAS The provider does not charge for entry tokens B The provider does not keep the prompts/responses, and ideally not the usable metadata; this reduces exposure but does not replace a possessed environment C The model automatically forgets its weights after each query D The application no longer needs internal logs

Question 14 — For a workload containing critical trade secrets, which option provides the strongest structural assurance?

HAS Standard API with contractual confidentiality clause B Cloud with ZDR, without further technical control C Owned and air-gapped environment, possibly with quantified open-weights model if self-hosting is required D Change supplier regularly to reduce dependency

Quiz — Session 8: Infrastructure & Deployment