AIMentor — AI Expertise Learning Platform

13 questions · pass mark 80% · 20 minute limit. The timer auto-submits at zero.

Time remaining: 20:00

Question 1easy
Why are containers especially valuable for AI apps?
They make the model more accurateThey package the app and its entire dependency-heavy environment so it runs identically everywhereThey reduce token costThey replace the need for a model provider
Question 2easy
What is the relationship between a Dockerfile, an image, and a container?
They are three names for the same thingA Dockerfile is the recipe; building it yields an image (immutable snapshot); running an image gives a container (live instance)A container builds an image which builds a DockerfileAn image is a running container
Question 3medium
Why copy requirements.txt and install dependencies BEFORE copying application code?
It is required syntaxEach instruction is a cached layer; installing deps first means editing code doesn't bust the (slow) dependency layerIt makes the app run faster at runtimeIt reduces the number of routes
Question 4medium
Which reduce Docker image size? (Select all that apply)
Using a slim/minimal base image (e.g. python:3.12-slim)A .dockerignore excluding .git, data, and cachesA multi-stage build that copies only finished artifacts into a clean runtime stageBaking API keys into the image
Select all that apply.
Question 5medium
What does a multi-stage build achieve?
It runs the app twice for reliabilityIt builds in one stage (with compilers/dev tools) and copies only the finished artifacts into a small runtime stage, so build bloat never shipsIt creates two containers per requestIt doubles the image size
Question 6medium
What does docker-compose provide?
A way to train modelsDefining and running multiple services together (app + vector DB + …) with wired networking, in one file/commandA vector similarity metricA prompt template engine
Question 7medium
In a compose file, how does the app service reach the database service named 'db'?
Via the host machine's public IPBy using 'db' as the hostname on the shared compose networkIt cannot reach itThrough the vector index
Question 8hard
Why must secrets like API keys come from the environment at runtime, not the Dockerfile?
It makes the image build fasterAn image is shared/pushed to a registry, so a baked-in key is a leaked key — and image history preserves it even if later removedThe Dockerfile can't hold stringsRuntime env is slower but safer
Question 9hard
How does env-based config support the LLMOps promote/rollback flow?
It forces a rebuild per environmentThe SAME immutable image runs in dev/staging/prod with only injected config differing — no rebuild to change config or roll backIt disables versioningIt embeds the environment in the image
Question 10hard
Which are containerization pitfalls? (Select all that apply)
Using a fat base image when a slim one would doCopying code before installing dependencies (busting the cache)Baking secrets into the imagePinning dependency versions in requirements.txt
Select all that apply.
Question 11easy
What is a .dockerignore file for?
Listing the app's routesExcluding files (.git, data, caches) from the build context so they don't bloat or leak into the imageConfiguring the vector databaseSetting the model temperature
Question 12medium
What Docker technique (two words) builds in one stage and copies only artifacts into a slim final stage?
Question 13hard
Which twelve-factor principle says configuration and secrets should come from where (one word)?

0 / 13 answered

Containerization — Final Exam