Software development kits (SDKs) have become deeply embedded within large-scale Python ecosystems. Teams often bundle open source projects, add some custom enterprise logic and publish it out to an internal package registry for usage across applications. While intended to accelerate development, the resulting dependency can create friction for an LLM when it can't understand. As a result, internal SDKs can see a loss in efficiency across maintenance and inheritance tasks, limiting the potential of coding tools. In this session, we will discuss strategies for optimizing dependency management and maximizing code understandability in Python, including:
- Designing Transparent SDKs
- Leveraging contextual instructions within parent projects
- Establishing open source contributions as the most logical solution