Part 11 · Senior Prep · Intermediate
Mistake Story Interview Q&A
Model STAR answers on verification mistakes, escape postmortems, personal accountability, and systemic fixes that prevent recurrence.
Failure ownership stories
Q: Describe a verification mistake you made
diagram
[INT][SENIOR][UVM] MODEL ANSWER
Q: Your mistake?
A:
STORY: Sampled coverage on driver stream — bins looked green, chip tape-out
review found monitor path never hit error bin; escaped scenario.
FIX: Moved subscriber to monitor; re-merged, found 12% func hole delta.
LEARNING: Closure metrics require DUT-visible sampling — now in team checklist.
SENIOR SIGNAL: owns mistake, shows systemic fix — not one-off luck.
PITFALL: 'I have no mistakes' — instant credibility loss.Q: Tell me about a bug that escaped to silicon
diagram
[INT][SENIOR][UVM] MODEL ANSWER (STAR)
Q: Silicon escape?
A:
SITUATION: W1C status register sticky-clear corner — passed block TB.
TASK: Lead postmortem without blame-shifting to design.
ACTION: Root cause: predictor modeled plain write, not W1C; added field semantic
test, plan row FP-33, predictor fix, re-merge closure.
RESULT: No repeat escape class in next chip — checklist item 'field semantics'.
OWNERSHIP: 'I signed off predictor without spec clause review' — personal.accountability.systemverilog
// mistake class: wrong predictor semantics — cite in story
function void write_status(uvm_reg_item rw);
// WRONG (my mistake): mirror = rw.value;
// RIGHT: model W1C — write-1 clears, read returns sticky
bit [31:0] wdata = rw.value[31:0];
mirror = (mirror & ~wdata) | (sticky_bits & mirror);
endfunctionQ: You signed off with a waiver that was wrong — what happened?
diagram
[INT][SENIOR][UVM] MODEL ANSWER
Q: Bad waiver story?
A:
STORY: Accepted design verbal on unreachable cross — no spec cite in waiver.
RESULT: Corner reproduced in emulation month later — firmware workaround required.
FIX: Waiver template now requires spec cite + design email + verification lead.
LEARNING: I rushed sign-off under schedule — process fix prevents repeat.
PITFALL: Blaming design for 'changing mind' — waiver process was mine to enforce.Key takeaways
Own a real mistake — coverage sampling, predictor semantics, bad waiver.
Escape postmortem → plan update + checklist — systemic fix required.
'No mistakes' answer fails instantly — prepare two honest stories.
Common pitfalls
Mistake story that blames tools without personal learning.
Escape story with no process change — sounds like bad luck only.
Recovery and integrity stories
Q: You discovered your own sign-off error one week before tape-out
diagram
[INT][SENIOR][UVM] MODEL ANSWER
Q: Late self-discovered error?
A:
ACTION: Immediate disclosure to lead + design — no quiet fix hoping nobody notices.
PLAN: Hotfix predictor, rerun closure subset overnight, update dashboard.
RESULT: Tape-out delayed 2 days — accepted because honesty preserved trust.
SENIOR SIGNAL: integrity over schedule cosmetic — teams remember who hid vs who flagged.Q: Conflict between schedule and correctness — time you chose correctness?
diagram
[INT][SENIOR][UVM] MODEL ANSWER
Q: Chose correctness over schedule?
A:
SITUATION: Pressure to waive last P0 cross — I had not run directed repro yet.
ACTION: Ran overnight directed — bin reachable; closure test added instead of waive.
RESULT: Found real RTL bug — schedule slip 3 days, saved respin cost.
SENIOR SIGNAL: short slip beats silent waiver that becomes respin.Q: How do you discuss failures in performance review self-assessment?
diagram
[INT][SENIOR][UVM] MODEL ANSWER
Q: Self-assessment failures?
A:
FORMAT: Mistake → impact → fix → metric improved.
EXAMPLE: 'Driver coverage sampling escape → added monitor-only rule → zero
sampling escapes next project (audit).'
MOTIVATION: Reviewers reward growth trajectory — not perfection claim.
PITFALL: Only listing successes — sounds unaware or dishonest.diagram
[INT][SENIOR][UVM] mistake story prep worksheet
Story A (technical): coverage / predictor / scoreboard class
Story B (process): waiver / sign-off / triage class
Each needs: personal role, fix, checklist/process change, measurable outcome
Practice: 90 sec STAR — timer on phoneKey takeaways
Late error discovery → disclose immediately — trust beats cosmetic schedule.
Correctness vs schedule — short slip beats silent bad waiver.
Self-assessment includes failures with metrics improved — shows growth.
Common pitfalls
Fabricated mistake too perfect — interviewers probe for depth.
Story without 'what I personally did wrong' — sounds like team failure only.