Hack Night - Can Your Agent-Written Code Survive Production?
- WHO IT IS FOR: Senior engineers, AI product builders, and technical founders focused on software reliability and agent-assisted development.
- WHAT IT IS ABOUT: A technical challenge testing whether code generated by AI agents like Claude Code and Codex can withstand real-world production failure modes.
- FORMAT: A hands-on build sprint and reliability challenge featuring a technical briefing, coding competition, and networking.
Move beyond the 'happy path' of AI-generated software at this intensive build night hosted by the [SF] Builders Collective, Reboot, and HackerSquad. While coding agents can create impressive demos, production environments are fraught with edge cases that often lead to critical failures. This event challenges engineers to put agent-written code under pressure, focusing on five specific failure modes including race conditions, timeout lies, and payment errors.
Participants will use tools like Claude Code and Codex to build applications and submit a proof package demonstrating how they identified and mitigated common AI-generated bugs. This is a high-stakes reliability challenge designed for skeptical builders and infrastructure experts who want to ensure their agent-assisted products are truly production-ready.
The evening kicks off with dinner and a technical framing session, followed by a competitive build sprint. Top performers will compete for prizes totaling up to $1,000 in cash and credits. Wrap up the night with PR-style reviews, live demos, and networking with San Francisco's top engineering talent.