Maria Bermudez is a developer educator and a lead developer of Docs AI Buddy, a documentation-assistance system built at Twilio to improve editorial quality and accessibility while keeping publication decisions in human hands.
Her work with Twilio spans communications infrastructure, developer tooling, and technical documentation. In 2020, she coauthored a guide to connecting WhatsApp with Twilio Flex. The following year, she helped explain how Pulumi and the Twilio CLI support infrastructure as code, including staging environments and automated production deployments. In 2022, she covered Flex Conversations and asynchronous customer messaging.
At AI Engineer World’s Fair 2025, Bermudez and Elmer Thomas introduced focused documentation agents. Bermudez demonstrated editing webpages and Markdown files, generating search metadata, writing image descriptions, and simplifying jargon. Proposed edits appeared as diffs with explanations tied to editorial guidance; she also acknowledged inconsistent results and described agent coordination as unfinished.
Her most distinctive contributions include:
- Human-reviewed accessibility automation: Bermudez built an image-description workflow combining structured prompts, batch processing, editorial review, and updates to documentation files. By October 2025, it had processed 2,734 documentation images, with every generated description reviewed by a person.
- Inspectable editorial assistance: Suggested revisions remain visible, explainable, and editable before entering pull requests or published documentation.
- Reusable documentation prompts: The public Docs AI Buddy prompt library shares templates for editing, accessibility, jargon reduction, and metadata. The prompts are open; the complete internal application is not.
