API Workflows & Tooling
Calling APIs, function-calling, and embedding pipelines.
Module Overview
Practical instruction for calling model APIs, handling structured outputs, error handling, retries, rate-limits, and building small production-like workflows. Includes function-calling and schema validation techniques.
Learning Objectives
- Implement reliable API calls and parse model outputs into structured formats.
- Handle API errors, rate limits, retries, and logging for reproducibility.
- Use function-calling or structured output prompts to integrate LLMs into apps or scripts.
Lesson-by-Lesson Breakdown
API basics: authentication, request/response lifecycle, environment variables.
Parsing outputs: strong parsing patterns for JSON, CSV, and delimited text.
Error handling patterns: transient errors, exponential backoff, idempotency.
Function-calling concepts and integrating external tools.
Logging and observability for debugging prompt pipelines.
Packaging a small script into a reusable CLI or web endpoint.
Hands-on Activities & Deliverables
Activities
Build a small integration: LLM tags a batch of text → structured JSON output → written to a CSV/DB. Provide a runbook explaining environment variables and how to reproduce results.
📦 Deliverable
GitHub repo containing code, sample inputs/outputs, and a runbook.
Required Tools & Readings
API docs (model provider), simple SDK examples, linting/checkstyle suggestions.
Assessment & Rubric
- Working integration & reproducibility50%
- Robustness and error handling25%
- Documentation & security25%
Prerequisites
Basic scripting knowledge (Python/JS recommended) and Modules 1–3 concepts.
Parent-Friendly Value
Students will be able to produce tools that generate structured, repeatable results for classroom or administrative use.