DeepSeek (and before it became DeepSeek)
What if I told you one of the most exciting new players in the competitive field of large language model wasn't founded by a Stanford or MIT AI PhD straight out of a top lab, but by a quant trader?Was this quant trader (LIANG Wenfeng, LWF) “the man who solved the Chinese market”? To what extent was his quant trading business develop within China?Why did LWF pivot from quant trading to LLM development?And the most mysterious of all, how was LWF (the founder of DeepSeek) able to compete among giants in and out of China, while all the other LLM giants are much better equipped than him?In this episode of Ascent, we're diving into DeepSeek, but not just any AI LLM story, we will be going through the founder’s journey, how he embarked on his first half of the career as a quant trader and creating a quant trading firm, and how did that quant trading firm later transitioned into an LLM company, which led to product we all know today - DeepSeek. Join us as we unpack this fascinating transition.Tune in to DeepSeek’s story!00:00 Intro04:23 Liang Wenfeng's early life and education08:16 Inception of the quant trading journey18:47 Becoming quant king: building a billion-bollar empire31:00 The hoarding (of GPUs) begins39:53 Liang Wenfeng's vision for China's quant future46:48 The 2021 challenge and fund drawdown50:29 The pivot: from trading to AGI57:17 Innovation under constraint01:08:31 DeepSeek's unconventional hiring philosophy01:12:35 Future uncertain: can DeepSeek outlast the giants?01:20:50 Ascent with open sourceCorrection: 59:24 - it should be 600 billion not 6 billion parameters(And guess who generated the show notes this time ... DeepSeek!)