AI 课程把 Jev 纳入教学并复盘

Arpit Bhayani @arpit_bhayani原推发布于 2026.09.27

在 X 上打开原推

My Applied AI Cohort - 5 wrapped up yesterday evening, and it was so much fun :) I will remember this cohort for two things - covering Jev, and this was the cohort during which px0 came to life :) We ended up covering Jev and a few other things that were not part of the original plan. A lot of things changed on the fly based on the discussions and the problems people brought into the room. What will stick with me is how many people told me that the things we covered were so fundamental that they would help them build and design almost any AI system. One thing I consciously optimize for in this cohort is practicality. We do not spend time on AI internals that you will never use. Instead, we focus on fundamental ideas, brainstorm around real problems, and cover techniques you can actually apply while building AI systems. The brainstorming was probably the best part. A lot of people came in with existing AI systems, and the ideas we discussed gave them ways to rethink, redesign, and improve what they had already built. That is exactly what I wanted this cohort to be - a set of mental models and engineering ideas that stay useful even as the models and tools keep changing. The discussions, as always, were the best part. Since almost everyone was building AI systems, we could go deep into real production problems and learn from each other's experiences. It went so well that I have decided to make this cohort part of the recordings as well. I am not sure when the next cohort will happen. Until then, I am going to focus on building px0. Best cohort to date. Really grateful to everyone who showed up, asked difficult questions, and made this cohort what it was :) http://arpitbhayani.me/ai