Generating Engineering Documentation from Code Modifications with LLMs

Presentation
Software Documentation

Anna Goncharova

Integration of GitHub Copilot and Gemini LLM into the engineering documentation workflow to support the automatic generation of accurate, context-aware documentation based on code changes in GitHub pull requests. The AI models analyze code updates and provide structured suggestions directly within pull requests, enabling engineers to contribute high-quality documentation more efficiently and without loss of coverage. The presentation includes implementation details such as prompt design aligned with internal documentation standards and writing guidelines, as well as integration of the solution into the CI pipeline.

Takeaways

The key takeaways offer practical guidance on integrating LLMs into Docs-as-Code workflows, generating documentation from code changes, and improving internal documentation practices.

Speaker

Anna Goncharova

Show profile

Biography

Name: Anna Goncharova

Experience: with six years of experience as a senior tech writer, I have a strong foundation in delivering exceptional technical documentation.

Expertise: I specialize in creating comprehensive, user-centric, and effective documentation for complex software products. My expertise spans various domains, including cloud computing, cybersecurity, and software development.

Contributions: as a Staff Documentation Engineer, I have made significant contributions to enhancing documentation practices. By collaborating closely with cross-functional teams, including engineers, product managers, and UX designers, I ensure that the documentation aligns with the product vision and meets user requirements.

Technical Skills: proficient in a wide range of documentation tools and platforms, such as MadCap Flare, Confluence, Jira, and Git. Experienced in version control systems and Agile methodologies, ensuring efficient collaboration and streamlined documentation processes.

Approach: I adopt a continuous documentation writing, "docs-as-code" approach, and AI-driven documentation writing for maintaining of tech documentation.

Passion: I am passionate about empowering IT engineers with the knowledge they need to effectively navigate and utilize tech documentation and data. I believe that well-crafted documentation plays a pivotal role in enhancing internal sharing of valuable knowledge, speeding up onboarding, user experiences and reducing support requests.

Back