@GenAI_is_real: this interview is worth rereading two years later. before yang zhilin founded moonshot, i attended his lectures for ove…

this interview is worth rereading two years later. before yang zhilin founded moonshot, i attended his lectures for over a year and we talked often. one thing he emphasized back then has stayed with me: engineering capability would only become more important over time, not less. in an era when everyone was chasing model architectures and paper counts, that was a genuinely non-consensus view. today, watching kimi push open source frontiers while the industry realizes that infrastructure and evaluation rigor are the real differentiators, his early judgment looks prescient. his guidance on evaluation and engineering discipline still shapes how i approach my work today. grateful to have learned from him before the world knew his name @zhang_benita