Senior / Staff · 45–60 min

System design interview
evaluation rubric

Objective scoring and architectural evidence — not binary right/wrong answers. Measures maturity, trade-off analysis, and system judgment for Senior (L5) and Staff (L6) roles.

Use with Stage 2 domain rubrics · CTO Talent Engineering guide

1
Needs improvement — below senior baseline
2
Mixed / borderline — mid-level depth, not senior bar
3
Meets senior bar — clear system judgment
4
Exceeds — Staff bar; edge cases and cost proactively

Candidate & interview context

Fields are for your notes only — nothing is submitted to Recberry.

Core evaluation matrix

Assign a score from 1 to 4 on each competency. Capture behavioural indicators and architectural evidence in the notes block under each section.

Competency 1 · 15% weight

Scoping, requirements & capacity estimation

Clarifying ambiguous problems, bounding constraints, and using back-of-the-envelope calculations to drive architecture.

  • Jumps into schemas or boxes without clarifying questions; ignores scale and functional constraints.

  • Basic functional questions but no QPS, storage growth, or read/write ratios.

  • Separates functional vs. non-functional requirements; reasonable sizing for memory, bandwidth, storage.

  • Metrics become architecture drivers (e.g. at 50k write QPS, relational transactions bottleneck — append-only log or queue).

Evidence / interviewer notes

Score assigned: _____ / 4

Competency 2 · 25% weight

High-level architecture & component design

Clean decomposition into decoupled services and data flows.

  • Monolith + single database; cannot explain flows between components.

  • LB → API → DB → cache pattern but tight coupling or obvious single points of failure.

  • Decoupled placement of gateways, brokers, workers, caches matched to access patterns.

  • Compares sync vs. event-driven options and trade-offs before recommending a path.

Evidence / interviewer notes

Score assigned: _____ / 4

Competency 3 · 25% weight

Data architecture, storage & consistency

Schema design, SQL vs. NoSQL choice, indexing, and consistency models.

  • Arbitrary database choice; no data model or access keys.

  • Reasonable DB types but weak on indexing, sharding keys, or isolation levels under pressure.

  • Choices tied to access patterns; explicit CAP / consistency vs. availability discussion.

  • Replication, partitions, idempotency keys, race conditions, eventual consistency lag.

Evidence / interviewer notes

Score assigned: _____ / 4

Competency 4 · 25% weight

Reliability, scalability & resiliency

Traffic surges, failures, partitions, and security under load.

  • Assumes 100% reliable infrastructure; no failure handling.

  • Mentions autoscaling but not connection pools, cache stampedes, or concrete mitigations.

  • Circuit breakers, rate limits, retries with jitter, DLQs, graceful degradation.

  • Zero-downtime deploys, multi-region failover, DR, encryption, zero-trust auth.

Evidence / interviewer notes

Score assigned: _____ / 4

Competency 5 · 10% weight

Communication, trade-offs & pragmatism

Technical communication, feedback, and business/cost constraints.

  • Dogmatic or defensive; buzzwords without mechanics; over-engineering.

  • Decent communication but interviewer must surface flaws and trade-offs.

  • Clear trade-off articulation; constructive response to pushback; effective diagram use.

  • Business empathy — engineering effort, ops overhead, cloud cost optimisation.

Evidence / interviewer notes

Score assigned: _____ / 4

Final decision & recommendation

Overall rating summary

Weighted scoreMultiply score × weight; sum for total out of 4.0

1. Requirements & estimation (15%) —
2. High-level architecture (25%) —
3. Data architecture & consistency (25%) —
4. Reliability & scalability (25%) —
5. Trade-offs & communication (10%) —
Total weighted score — / 4.0

Recommendation outcome

  • Strong hire (3.5 – 4.0) — Staff-level vision, deep trade-offs, proactive fault tolerance.

  • Hire (2.8 – 3.4) — Solid senior; meets core architecture with minor coaching on edge cases.

  • Borderline / downlevel (2.0 – 2.7) — Consider mid-level (L4) if potential without senior depth.

  • No hire (< 2.0) — Cannot design scalable components or justify design choices.

Key strengths

—

Areas for improvement / concerns

—

Need help standardising interviews?

We design domain rubrics and hiring infrastructure your team owns — aligned with how senior engineers are actually evaluated.

© 2026 Recberry · Internal evaluation template — share with interview panel as needed