ok so i've been researching this obsessively because i'm applying for Perplexity's new grad SWE roles in 2026 and there is almost nothing written from a new grad perspective. everything is L5/senior.
here's what i've pieced together from various forums, DMs, and one friend who got through to final rounds:
the process for new grads seems to be: OA (online assessment): medium-difficulty leetcode style, 2 problems in 90 minutes, standard stuff technical phone screen: 1 coding problem, 45 min, they ask you to explain your reasoning as you go virtual onsite: 2-3 rounds. at least one is more coding, one is a system design-lite (they scale it for experience, so not "design Google" but more "design a simple rate limiter" or "how would you model search results"), and one is behavioral
what my friend said they focused on for Perplexity specifically: they asked about AI/LLM concepts at a high level. not deep ML math, but like, what is a token, what's retrieval-augmented generation, why does latency matter in search. if you're applying to a company whose product IS an AI answer engine, you should understand how it works at a surface level. the behavioral was surprisingly real. not "tell me a time you showed leadership" canned stuff. more like "describe a project that didn't go as planned and how you handled it." detail matters.
my prep plan so far: grinding leetcode mediums (no hard yet, aiming for 120 mediums before i apply) reading about RAG and vector search at a conceptual level doing a few mock system design questions at junior-appropriate scope
if anyone's been through the Perplexity new grad loop recently please drop what you remember. the more data points the better for everyone.