kpowerinfinity

Traditional narrative is that you need tons of traces of actual execution to post-train and get models right. Nobody seems to use Kimi API from Moonshot, I bet everybody is using them on neoclouds/inference providers like Together, Nebius, Fireworks etc. where unlikely they will get traces (in fact, thats the whole promise of these inf providers). How are Kimi models improving so quickly? Is this just distillation (though Sol/Fable just came out so I find it hard to believe)

An exploration of ways in which you can hold the bar for technical decision making as a company and engineering team scales.

OP here. We all know leadership behaviors that lead to success, but what do we need to watch out for? How do we really prioritize what activities to focus on for new managers? I was recently helping somebody through this question, and decided to pen my thoughts for a broader audience.

OP here. Always a good place to be in. Hard numbers is more valuable but you need to think about how to address the concerns else you will hit a glass ceiling, or the moment they find an alternative they will jump. Thinking about lock-in is very important.