AI systems rely on APIs for data and connectivity, and conversely, AI can be used to accelerate API development.
Running large language models locally for code assistance can reduce costs and improve data privacy compared to cloud-based services.
AI applications typically consist of a user interface that communicates with a model inference server via a standardized API, like the OpenAI API.
Using open source models from platforms like Hugging Face, such as IBM's Granite, ensures compliance and avoids potential licensing issues.
Configure a local environment by running a model with InstructLab and connecting it to a VS Code extension like Continue via its OpenAI-compatible API.
Use a local AI assistant integrated into an IDE to generate a complete OpenAPI specification from a simple natural language prompt.
Identify and correct issues in an AI-generated OpenAPI specification by using a linter like Spectral and then prompting the AI to make specific fixes.
Generate custom Spectral linting rules by providing a natural language prompt to the AI assistant, enforcing specific governance policies for your API.
Provide the OpenAPI specification as context to an AI assistant to generate the corresponding Java backend implementation using the Quarkus framework.
Use AI-generated code with caution as context can be limited, and leverage open source tools to experiment with these capabilities in your local environment.
31:39Rainer Stropek
32:26Krzysztof Cieślak
29:50Christopher Walles
23:55Zan Markan
35:16Ankit Patel
32:13Cedric Clyburn
58:06Anna Fritsch-Weninger
31:56Prashanth Chandrasekar, Alejandro Saucedo, Jakob von Lindern & Demetris Cheatham
Jobs that call for the skills explored in this talk.
webLyzard
Vienna, Austria