Season 5: The AI Proposal · Episode 2 of 8
The Model Audition
You play: AI Engineering Intern
Included with Core programme and Certification prep
The situation
Code Nexus Assist — the internal support assistant — is cleared to proceed. Now it needs a model. Maya has three candidates in the Foundry simulation, each with a model card. The constraints are firm: the assistant answers short support questions, it must run in the westeu region for data-residency reasons, responses need to come back fast, and the per-request cost has to stay low because support volume is high.
What you'll learn
- AI-901 model capabilities and implementation: reading a model card (context window, cost, regions, modalities), deployment as a state change, generation parameters like temperature, system vs user prompts, and writing a lightweight Python client against a model SDK.
- Tokens, context windows and max tokens; embeddings; the transformer and attention; temperature and top-p; large, small and multimodal models; and the Microsoft Foundry model catalog and playground.
Who you work with
Maya Chen
Senior AI Engineer · Your mentor
Priya Naidoo
Cloud Architect
Scenes
- 1.Three candidates
- 2.Read the cards
- 3.What does this setting do?
- 4.Write the client
- 5.Deploy it for real
- 6.Implementation check
What you leave with
You selected Assist Lite from fictional cards, implemented the teaching SDK client and tested a simulated low-temperature deployment. Its region and cost fit this fixture. Production quality, residency, authentication, failure handling and billing still require provider-specific review.