Andrew Orobator was introduced in his 2026 AI Engineer World’s Fair biography as a senior Android engineer at Reddit, with more than a decade of experience building Android products at scale across consumer apps, developer tooling and mobile architecture. He is the author of Vibe Engineering, a ten-part methodology for AI-assisted software development covering personas, reusable skills, work logs, agent workflows and self-driving codebases. He co-authored the series with Claude using the practices it describes, treating AI as a collaborative engineering system with memory, process and taste.
His work explores how coding agents can become durable engineering infrastructure that preserves context, improves through feedback and helps teams ship better software with less coordination overhead. In his World’s Fair presentation, “Scale the Judgment, Not the Model,” he argues that reliability depends on making engineering judgment explicit through reusable skills, work logs that carry decisions and progress across sessions, and personas that supply security, design and other review perspectives. He used a work log to develop the presentation across multiple sessions and describes requiring agents on his personal projects to record running application flows for his review. His verification ladder extends from builds and tests to screenshots, recordings and production telemetry, with humans retaining approval and merge authority.
Orobator also built a local feature-flag cleanup agent that screens candidates using code complexity and experiment data before passing safe mechanical changes to a model. He reports seven pull requests with passing CI at $1.26 each, leaving review and merging to him. Drawing on an agent’s unsolicited addition of an emergency-recovery exception to a repository safeguard, he advocates enforceable gates with operator-controlled bypasses. His practitioner’s approach combines narrow specialist agents with postmortem feedback and reviews of stale guidance to make AI-assisted development more reliable and useful.
Andrew Orobator shows how skills, work logs, review perspectives and enforceable gates turn coding agents into useful maintenance systems—and why keeping that judgment current becomes part of the engineering job.
Skills preserve reusable decisions, work logs preserve progress across sessions, and personas supply different review perspectives.
▶ Watch ↗