Beyond the Hype:LLMs aren't Magic

Abstract

This talk demystifies Large Language Models (LLMs) by presenting them not as magic, but as powerful APIs that developers can master using existing software engineering skills. It advocates for an engineering-based approach, demonstrating how to use Python's abstract base classes to create a vendor-agnostic and testable interface. This allows for interchangeable backends, from cloud services like Gemini to local mocks, making AI development more modular and robust. The talk empowers developers to look past the hype, and instead apply their existing knowledge to build reliable and innovative applications, concluding with the message that with LLMs, "It's just an API. It's just Python. Grab a key, write a test, and build the future."

Speakers