Applications of Foundation Models – AIF-C01 Study Guide
This is the largest domain (28% of the exam) and the most practical: how to actually build with foundation models. It covers prompt engineering, inference parameters, Retrieval-Augmented Generation, fine-tuning, agents, evaluation, and model selection. The highest-value skills are knowing when to use prompting vs RAG vs fine-tuning, and what the common inference parameters do. Learn those decision boundaries cold.
Prompt Engineering
Techniques to shape model behavior at request time (no retraining):
| Technique | What it is | When to use |
|---|---|---|
| Zero-shot | Instruction only, no examples | Simple tasks the model already handles |
| Few-shot | Include a few worked examples showing the pattern | Steer format/style with demonstrations |
| Chain-of-thought | Ask the model to reason step by step before answering | Complex reasoning/multi-step problems |
Prompt engineering is the cheapest lever — it changes inputs, not the model's weights.
Inference Parameters
| Parameter | Controls | Effect |
|---|---|---|
| Temperature | Randomness/creativity | Higher = more varied; lower (→0) = focused, deterministic |
| Top-p (nucleus) | Diversity of token choices | Narrows or widens the sampled token set |
| Maximum tokens | Output length | Caps how many tokens the model generates |
For factual/compliance tasks, lower the temperature. To bound length or cost, set max tokens. Temperature and top-p shape randomness; max tokens shapes length — don't confuse them.
Unlock with a paid plan
5 more sections (about 395 words) continue below. Every paid plan unlocks all AIF-C01 study guides, the full question bank and mock exams, from $4.97/month.