CompleteEvent Demo

Claude API Fundamentals & Prompt Engineering

4:00 PM – 6:00 PM

Start with the core API surface: messages, roles, tokens, and context windows. Work through progressively complex prompts and learn how model behavior shifts with system prompt design. You'll leave with a clear mental model for reasoning about latency, cost, and quality trade-offs. Hands-on: build a basic chat interface from scratch using the SDK.

Tool Use & Function Calling

6:15 PM – 7:30 PM

Deep-dive into Anthropic's tool use API. Learn how to define tools, handle multi-step tool call loops, and design safe, deterministic tool schemas. We'll build a real agent that fetches weather data, queries a database, and sends emails — all orchestrated by Claude. Hands-on: build and test a multi-tool agent end-to-end.

Building a RAG Pipeline from Scratch

8:30 PM – 10:30 PM

Retrieval-augmented generation demystified. Cover the full pipeline: chunking strategies, embedding models, vector store options (pgvector, Pinecone, Weaviate), retrieval scoring, and re-ranking. Walk through the architectural decisions that separate a quick demo from a production system. Hands-on: build a document Q&A tool over a custom knowledge base.

Streaming Responses & Real-Time UX

10:45 PM – 12:00 AM

How to stream tokens from Claude to your frontend and why it matters for perceived performance. Covers server-sent events, React streaming patterns with Next.js App Router, skeleton loading, and how to handle tool call interlacing in streamed responses. Hands-on: add streaming to the chat interface you built in session 1.

Testing & Evaluating LLM Outputs

12:15 AM – 1:00 AM

The discipline most teams skip. Learn how to write deterministic unit tests for LLM features, build eval harnesses that catch regressions, use Claude-as-judge patterns, and instrument your app for production monitoring. We'll cover Braintrust, PromptFoo, and a hand-rolled baseline. Hands-on: build an eval suite for the RAG pipeline from session 3.