In the LLM space, a few things I'm interested and are valuable - 

* Inference
* Teach models to do things better (training/RL)
* Agent / harness (e2e and system programming (container, tooling))


Inference is what I've been doing. it's hard and hard-core. not sure if I
really love it. i want to give myself some more time, mindset should be to
fully embrace and absorb all the learnings. This is a super deep area and
requires a lot of focus and dedication. Not my strong suite. I have a lot of
breadth, passionate about tooling.