Train AI Supervisors,
Not AI Consumers

Master the art of AI supervision through hands-on coding challenges. Progress from local models to Claude Sonnet by proving your capability with weaker models first—inspired by AI alignment research.

The Weak-to-Strong Learning Path

Close the "AI literacy gap" by learning to steer more capable AI through precision, verification, and intent alignment.

Local Models First

Start with local SLMs (Llama 3.2) to build fundamentals without dependency

Prove Capability

80% success rate unlocks stronger models—earn access through competence

Deploy Projects

Build portfolio-ready projects across Web, Data Science, and Cloud tracks

Why Weak-to-Strong?

Unlike generic coding platforms, we focus on AI supervision skills

Anti-Blind-Prompting

Explain your approach before coding, then explain your output after. No "just build it" allowed—develop intentional AI interaction patterns.

Progressive Model Access

Earn Claude Haiku by mastering local models, then unlock Sonnet through demonstrated competence—not subscription tiers.

Red Team Checkpoints

Security challenges embedded in the curriculum teach you to spot XSS, SQL injection, and PII exposure in AI-generated code.

Portfolio-Ready Output

Three specialized tracks produce real deployable projects: responsive web apps, data pipelines, and cloud infrastructure.

Ready to Master AI Supervision?

Join the beta and help shape the future of AI literacy training

🚀 Beta Access • Free During Development