Was speaking to a friend starting a Data Science master's and he asked me this question.

For content, currently working at FAANG, here's what I see:

1. Sharing of code is a pain. Keeping code updated is a pain. I've only been on a single team that uses Git in a decade. And even then, it wasn't really used.

2. Data quality is awful. We are constantly working with other teams to get it fixed, no one really cares but we have to deal with the consequences.

3. Most people want to work on cool LLM projects but there are too many internal blockers. (We don't have access to openAI APIs, we don't have good GPUs to run models in our dev environment, the few models that run on our underpowered GPUs are bad)

4. Thankfully not a problem for me but the hiring market is insane right now.