I went through Meta's full interview loop for an E5/E6 Software Engineer role. The process included the standard four-round onsite: two coding rounds, a system design round, a behavioral round, and an AI coding round that Meta has added to their loop. I prepared extensively for each stage — grinding LeetCode-style problems for the coding rounds, building a library of system design references, and writing out STAR-format behavioral answers calibrated to Meta's E5/E6 expectations, drawing on my Oracle Cloud Infrastructure work across the Multicloud Observability team (control plane unification, data plane migration to Oracle Managed Kubernetes, and the Oracle Database at AWS buildout).
Ultimately, I received a rejection with a one-year cooldown before I can reapply.
Looking back honestly, a few weaknesses stood out:
Coding execution under time pressure. While I could solve the problems, I wasn't always optimal on the first pass. I spent time re-deriving approaches instead of pattern-matching quickly, which cost me on the second problem in at least one round.
System design depth vs. breadth tradeoff. My background is deep in cloud infrastructure and observability, so when the design prompt pulled toward consumer-scale product systems (feed ranking, social graph type problems), I leaned on general principles rather than Meta-specific intuition. I covered the fundamentals but didn't always drive the conversation into the nuanced tradeoffs interviewers wanted to hear.
Behavioral calibration to Meta scale. My STAR stories were strong on technical substance, but a few of my impact framings were sized for Oracle's context rather than translated into the scale and cross-org influence language Meta's bar expects at E5/E6.
AI coding round unfamiliarity. This was a newer format for me and I hadn't practiced it as deliberately as the traditional rounds, so my workflow with the AI tooling wasn't as fluid as it could have been.
The cooldown gives me a clear runway to address each of these before reapplying.
