Post-Training Foundation Models: Environments, Rewards, and Verifiers

AI labs are pouring money into RL environments. Anthropic alone has discussed spending over a billion dollars on them in a single year, and a whole class of startups now exists to sell labs the worlds their models train in. Here's the part that matters for you: the skill they're buying is not "run PPO." It's designing the tasks, tools, rewards, and verifiers that make RL post-training work. That skill is software engineering plus evaluation design. No PhD required. I build this layer for a living, and in four weeks I'll teach you to build it too. One more thing before you scroll to the schedule. You will not build eight disconnected demos in this course. You'll build one environment that compounds: created in week 2, hardened against reward hacking that same week, used to post-train a real model in week 3, evaluated with statistics that survive scrutiny in week 3, and extended to a team of collaborating agents in week 4. By the showcase, you'll have shipped an environment you can publish to Prime Intellect's Environments Hub and point to in an interview. Not a certificate. An artifact.