Perplexity · Primly Community

How much AI/LLM knowledge do they actually expect for SWE roles?

staff_steph · 2 replies

I have a Perplexity SWE screen next week and I'm kind of spiraling. My background is backend infra, some distributed systems, zero formal ML training. I've used LLMs but I don't have a deep foundation in transformers or anything.

Is this disqualifying? The JD didn't say ML required but it's Perplexity so I assumed... are they really testing on model internals or is it more systems-adjacent stuff? Really don't want to go in unprepared.

2 replies

ml_mike

for pure SWE roles (not ML eng) it's more systems than model internals in my experience. you should know roughly how inference pipelines work, why latency matters for LLM serving, what chunking/embedding means at a high level. you do NOT need to explain backprop.

tired_recruiter

recruiter hat on: SWE roles at AI companies vary a LOT by team. ask your recruiter which team you're interviewing for and then ask specifically if the role requires ML depth. most won't dodge the question.