Just watched Why Not TV's @TairanHe99. interview with OpenAI's post-training lead Weng Jiayi. @Trinkle23897

Here are some thoughts 🧵

1️⃣ Impact
The discussion on influence & evaluation systems struck me. Weng's "impact" isn't just metrics—it's about being genuinely helpful to those around you. Real influence comes from creating positive externalities while becoming a better person.

2️⃣ Greatness emerges
ChatGPT's breakthrough wasn't planned. The RL team (including PPO's John Schulman & Meta's new chief scientist Zhao Shengjia) didn't expect much from their 12k-user prototype. Shows how organizational structure matters—great algorithms die without proper engineering support.

3️⃣ Choices define us
Weng had offers from DeepSeek's predecessor, NVIDIA, and Google. He chose OpenAI's RL team. At 24, he couldn't have known this decision would let him contribute to the AI renaissance. Timing + effort = opportunity.

4️⃣ Future outlook
Counter-intuitive insight: Engineers transitioning to research is easier than researchers learning engineering. If true, we're entering the engineer's era. The path to AGI needs both Boris Cherny @bcherny (Claude Code) and Harrison Chase (Langchain) types—everyone shines in their own way.