Interview prompt
Problem context
Skills being evaluated
Use the sequence below to surface constraints, choose boundaries, test failure behavior, and defend trade-offs. Concrete numbers are interview assumptions, not claims about a real production system.
Clarify the decision
- Define debt as a present constraint or risk with evidence, not age or aesthetic dislike. Identify strategic goals, delivery bottlenecks, incident exposure, compliance dates, and irreversible decisions.
Establish scale assumptions
- Quantify recurring engineer time, change lead time, incident loss, cloud spend, blocked revenue, and probability-weighted risk. Use ranges and confidence, not fabricated precision.
Functional and non-functional requirements
- The portfolio compares unlike work transparently, preserves urgent risk, funds enabling foundations, and includes stop or reassess gates. Owners must remove temporary migration assets.
High-level architecture
- Maintain a debt register linked to business journeys, system owners, evidence, risk, dependency, intervention options, and outcome measures. Group work into themes and sequence enabling constraints before dependent migrations.
Data model and flow
- Signals from incidents, delivery metrics, cost, audits, and roadmaps feed quarterly review. Each initiative has baseline, target, leading indicator, decision date, and exit criteria.
Consistency and transaction boundaries
- Use a common scoring frame but preserve qualitative legal and existential risks that should not be averaged away. Reassess as product direction and evidence change.
Failure modes and recovery
- Avoid one massive program by staging options and checkpoints. If benefits do not appear, stop, narrow, or change approach rather than protecting sunk cost.
Security and privacy
- Security and compliance debt includes exposure, exploitability, compensating controls, and deadlines; it is not automatically top priority without threat context, nor optional because benefit is hard to monetize.
Observability and SLOs
- Track constraint removal, incident recurrence, lead time, cost units, exception count, migration progress, and decommissioned assets. Service count or lines rewritten are not outcomes.
Capacity and cost
- Compare do-nothing cost, partial mitigation, incremental modernization, and replacement. Fund options that cheaply reduce uncertainty before committing to expensive paths.
Alternatives and trade-offs
- A single score simplifies ranking but hides dependencies and fat-tail risk. Use scoring to structure debate, then make accountable portfolio decisions with documented assumptions.
Evolution and migration
- Pilot the register on one product area, calibrate estimates against delivered work, and connect it to planning. Retire items when the constraint disappears, even if the technology remains old.
What Staff and Principal candidates should emphasize
- Principal candidates turn architecture into capital allocation. They quantify enough to decide, keep uncertainty visible, and create reversible stages rather than lobbying for favorite technology.
Decision trade-offs
Prioritization
Option A
Rank by a single numeric debt score
Option B
Portfolio review with common evidence and risk classes
Recommendation:Use common evidence for comparability but retain dependencies, fat-tail risk, and strategic option value in accountable review.
Intervention
Option A
Full replacement of the indebted system
Option B
Targeted constraint removal and staged options
Recommendation:Choose the smallest intervention that removes the binding constraint, then reassess before funding further modernization.
Follow-up interview questions
- 01How do you compare compliance risk with developer toil?
- 02When is an old technology not debt?
- 03What makes a rewrite an option rather than a commitment?
- 04How do you stop a failing architecture program?
Common weak answers and mistakes
- 01Calling old or unfashionable technology debt without a measurable constraint.
- 02Using a precise score to disguise uncertain assumptions.
- 03Funding a rewrite without intermediate value or stop gates.
- 04Measuring migration output rather than removed business and operational constraints.
Interviewer evaluation rubric
Creates a technology cleanup list without business evidence, risk, sequencing, or outcome measures.
Builds a register with owners, evidence, risk, cost, options, dependencies, and measurable targets.
Uses ranges, option value, do-nothing comparisons, staged gates, and portfolio governance across risk classes.
Changes executive capital allocation through credible architecture evidence while protecting uncertainty and organizational learning.