Backend screening is about scale and failure. State the traffic you handled and what broke — vague 'high concurrency' claims get discounted.
Three things, in order of weight.
Real numbers: QPS, data volume, P99 latency, error budget
A production incident you handled and what changed afterwards
Design trade-offs — why this storage, why this consistency level
Work them into your title, project descriptions and skills — don't pile them into a list.
Each cover is the template itself, rendered.
Single column, parses cleanly and holds long bullets.
Heavier headings suit a systems-oriented history.
Pure black and white when the content is dense.
One page. Unless you have 10+ years and a genuinely full second page, the second page only dilutes the first. Put real numbers: qps, data volume, p99 latency, error budget above the fold.
'High concurrency, high availability' with no scale attached. It is the single most common reason this role gets screened out, and fixing it matters far more than changing templates.
Use AI for the first draft, then verify every claim and number yourself and reorder keywords against the target job description. AI handles speed; the content is still your responsibility.