Most developers encounter AI APIs through a tutorial that skips the parts that actually break in production.
This workshop addresses the gap between reading documentation and writing code that reliably calls an AI service, processes its output, and handles edge cases without crashing. We cover REST and streaming endpoints, token limits, rate limiting, and error handling patterns that apply across providers like OpenAI, Anthropic, and Google Gemini.
What the sessions cover
Each session is built around a concrete problem rather than a feature list. You will work through authentication flows, understand how API keys differ from OAuth tokens in AI contexts, and learn why naive retry logic can burn through your quota in minutes.
Structured outputs and prompt engineering
We spend time on structured outputs, JSON mode, and function calling, because these are the patterns that make AI responses actually usable in an application. Prompt engineering is treated as an engineering problem, not a creative exercise, with attention to token budgeting and determinism.
The workshop is designed for developers who already write code but have not worked directly with AI APIs before. Familiarity with HTTP requests and at least one scripting language is assumed.
- REST and streaming API patterns
- Authentication and key management
- Rate limits, quotas, and cost estimation
- Structured outputs and function calling
- Error handling and retry strategies