>As a quant, you can be doing ML to predict the stock market or hard core network coding to make placing orders faster.

I think he was pointing out that what is doing is not that relevant to advancing human knowledge. Proprietary trading strategies and optimizations that will most likely never be made public are not any better.

All of your examples are the same until you hit Google/FB research where you might actually be working on something that isn't directly promoting the bottom line of your employer.