The Meta E5 coding signal, deconstructed

The twenty-one-data-point survey

We collected twenty-one recent E5 onsite reports from candidates across the last eighteen months. Not a statistically significant sample, but enough to pattern-match.

What gets asked

Across two coding rounds, we see this distribution.

Graph problems: 43 percent of rounds. Mostly implicit graphs (grids, state-space BFS) or adjacency-list traversal. Dijkstra appears in maybe one in ten. Heavy graph theory rare.

Dynamic programming: 29 percent. Always medium, rarely hard. One-dimensional DP dominates (climb stairs variants, coin change, house robber variants). Two-dimensional DP on a grid appears in maybe one in five.

Array and string: 24 percent. Sliding window and hash-of-seen split this category roughly evenly. Low on-paper difficulty, but interviewers probe hard on complexity reasoning and edge cases.

Tree and recursion: 19 percent. Classic post-order aggregation (diameter, max path, LCA). Sometimes with a twist like path sum tracking.

Heap and priority-queue: 10 percent. Mostly top-k variants.

Backtracking: 5 percent. Rare at E5, more common at E6.

Total exceeds 100 because some rounds combined two patterns.

What interviewers weigh

Speed to first insight. At E5, interviewers expect the pattern recognized within two to three minutes. Beyond that, they mark you down on problem decomposition.

Running code. E5 bar is working code on the first submission for at least one round, working code after one correction for the other.

Complexity reasoning. You should state time and space complexity without being asked. At E5, getting one wrong is forgivable if you catch it during trace; getting one wrong and never noticing is not.

Edge case discipline. Empty input, single element, all-same elements, maximum input size, integer overflow if applicable. E5 interviewers test these with follow-ups. E6 interviewers expect you to raise them unprompted.

What differentiates E5 from E4 signal

E4 is "can you code medium-grade problems correctly." E5 is "can you code medium-grade problems quickly AND reason about the tradeoffs AND communicate the reasoning clearly AND handle a curveball follow-up without panicking."

The gap is communication and composure, not raw algorithm skill.

What differentiates E5 from E6 signal

E5 is expected to handle medium-hard problems cleanly. E6 is expected to handle hard problems, to raise meta-concerns (scalability, production-readiness of the algorithm), and to engage with the interviewer at a peer level during the session.

Preparation weighting we recommend

Spend forty percent of coding prep on graph and state-space BFS. Second biggest bucket is DP at thirty percent. Then array and string at twenty percent. Top off with tree, heap, and backtracking.

Above raw problem count, practice communicating while coding. Record yourself. The signal Meta E5 interviewers weight most is how clearly you narrate as you work.