> I know that if I leave my current company, it will be very hard to find a similar level of benefits

I have heard this dozens of times on this thread. I bet it's not true. I am full remote, senior level SWE, make \~$300-350k a year depending on stock price,, not factoring benefits. For most part good wlb, more recently less true but hopefully dies down a bit after next product launch. I'm 30 so presumably you have \~20 yoe on me and could get something better. If you can get director level or higher, there are definitely full remote jobs that would pay you a lot more.

I work on a LLM platform, and meet with vendors (think OpenAI, AWS, Google, Databricks) so I'd like to think I have an above average exposure to this area. There are really 3 very different verticals in AI.

1) Research, basically people working on foundation models, training, new model techniques, etc. This is what you hear with "zuck poaches research scientist X for $XX m". imo hard to break into unless you got lucky and this happened to be what you were working towards in a PhD program.

2) LLMops- basically extension of ML ops. Running LLM products at scale, making efficient use of LLMs, studying latency/throughput/etc (standard engineering metrics). This is what I work on, no specialized background kind of just lucked into it out of being at my company for years and we needing support in this area. Lot of generic platform engineering (eg. for data eng) can probably pivot into this.

3) LLM product- Product teams building products using LLMs, think Chatbot for company X, or insights extraction for product Y. Probably just more traditional product development background but there are some LLM-specific items like Langchain/Langgraph, MLFlow/Langsmith for LLM observability. probably easiest to pivot to from internal team, but maybe any strong product background could be a viable hire here.

I think you need to take some time to figure out what about LLM actually interests you beyond hype hype hype. Easier said than done of course