Uber interview

Uber Behavioral Question Bank

I – Strategy & Technical Vision

1. Long-term technical vision.
2. Balancing innovation vs delivery pressure.
3. Introducing new architectural paradigm.
4. Aligning tech direction with business goals.
5. Sunsetting / deprecating legacy systems.

II – Complex Problem Solving

6. Technically complex cross-team problem.
7. Disagreeing & influencing architecture outcome.
8. Design failure → recovery.
9. Leading mitigation during outage.
10. Large-scale optimization or efficiency gain.

III – Cross-Functional Influence & Alignment

11. Conflicting priorities between product & engineering.
12. Buy-in for unpopular technical investment.
13. Influencing executives / senior leadership.
14. Resolving inter-team misalignment.
15. Mapping technical work to business outcomes.

IV – Execution & Delivery Leadership

16. Rescuing off-track project.
17. Delivering amid scope / requirement changes.
18. Cutting scope / delaying launch.
19. Scaling team delivery via process / tooling.
20. Introducing metrics / observability that changed behavior.

V – Leadership & Mentorship

21. Mentoring senior / staff engineer.
22. Handling under-performance.
23. Building culture of technical excellence.
24. Managing up & communicating risk.
25. Giving feedback across seniority.

VI – Scalability & Reliability

26. Scaling system ×10.
27. Improving reliability / latency.
28. Proactively addressing systemic reliability issues.
29. Influencing org-wide reliability / incident mgmt.
30. Balancing SLO vs velocity.

VII – Decision-Making & Trade-Offs

31. Controversial technical decision & defense.
32. Refactor vs rebuild.
33. Decisions with incomplete data.
34. Choosing long-term scalability over short-term velocity.
35. Optimizing for learning over delivery.

VIII – Communication & Influence

36. Explaining strategy to non-technical audience.
37. Mediating senior-engineer disagreement.
38. Evangelizing best practice / tech.
39. Aligning distributed teams.
40. Managing architecture review / decision records.

---

Story Index

#	Title	Primary Theme	Uber Dimension(s)	Questions	Time (min)	Link
9	Claims Modernization – Cross-Team Leadership	Ambiguous multi-team execution	Execution / Leadership	6 14 16 30 39	4	Claims Modernization – Cross-Team Leadership
11	Claims Modernization – Legacy Decommission Strategy	System migration & risk management	Strategy / Reliability	5 6 11 28 29 30	4	Claims Modernization – Legacy Decommission Strategy
12	Claims Modernization – Milestone Delay & Recovery	Ownership & course correction	Execution / Decision	16 18 24	3	Claims Modernization – Milestone Delay & Recovery
13	Claims Modernization – Conflict Resolution Between Tech Leads	Conflict mediation & alignment	Collaboration / Influence	37 11 12	3	Claims Modernization – Conflict Resolution Between Tech Leads
3	Fraud Defense – Team Transformation	Upskilling & process improvement	Leadership / Execution	19 20 21 22 23 38	4	Fraud Defense – Team Transformation
14	Fraud Defense – Glue Pipeline Incident Recovery	Incident response & learning	Reliability / Learning	9 27 29	3	Fraud Defense – Glue Pipeline Incident Recovery

---

Capital One – Claims Modernization

Perfect — I understand exactly what you need:

✅ Keep the dense, tactical STAR++ note style (like “Legacy Decommission Strategy”). ✅ Integrate Jeff Sipe structure → Situation (when/where/timeline) · Task (simple + time pressure or scope) · A0–A4 (What / How / How matters) · Result (impact + scale + learning). ✅ Always show timeline / urgency / scope so the interviewer feels the weight of the role. ✅ Drive every story’s first Result line to the question theme and end with repeatability. ✅ Each story fits 2–3 min spoken (≈ 12–15 lines).

Because all 15 stories in one message would exceed the limit, I’ll send them in 4 compact batches (≈ 4 stories each). Below is Batch 1 – Capital One Claims Modernization (4 stories) in the upgraded style.

---

Claims Modernization – Cross-Team Leadership

S: Sep 2024 → Feb 2025, Dist. Eng. Capital One Fraud Tech. 16 contractor teams (≈250 eng) rushed after Discover deal; ambiguous ownership. T: Stabilize and deliver critical workflows under 6-month timeline without formal authority. A0 – Pre-Exec: Audited 5 cross-team handoffs + reviewed 70 PRs → identified testing / ownership gaps. A1 – Define Structure: Mapped dependencies, drafted RACI matrix shared with VPs. A2 – Uplevel Leads: Ran code reviews + showed Ports/Adapters pattern through live refactor. A3 – Quality Push: Introduced unit + contract tests mandatory before integration. A4 – Delivery: Flagged Dec delay 6 wk early → reset scope + coached teams for Feb launch. R: Feb milestone on time; first phase eliminated stuck claims ($34 M/yr efficiency). Patterns replicated in 2 later modernizations. TRAITS: Execution • Leadership • Alignment (Q 6 14 16 30 39)

---

Claims Modernization – Legacy Decommission Strategy

S: 2024 Q4 → Q1 2025. Legacy Java monolith (1 M claims ≈ $500Mexposure)causing10$50 M risk ). T: Within 3 months, migrate to Lambda Router + Step Functions w/out downtime or data loss. A1 – Map + Plan: Cataloged 5 resolution strategies; pin-pointed dual-write fault zones. A2 – Design Router: Workflow-agnostic routing policy by strategy; idempotent payloads + audit IDs. A3 – Pilot: Cut over 2 strategies (40 % volume); shadow diffs verified 0 state drift. A4 – Operationalize: Runbook + rollback templates → 16 teams trained in 2 weeks. R: Stuck-claim rate 0 % on migrated flows (~ $20 M risk removed); pattern became Fraud Tech standard. TRAITS: Strategy • Risk Mgmt • Reliability (Q 5 6 11 28 29 30)

---

Claims Modernization – Milestone Delay & Recovery

S: Joined Sep 2024 into project started Jun; first deadline (Dec) already at risk. T: Assess and communicate delay within 30 days → recover Feb release. A1 – Audit: Found over-granular Step Functions (>30 states) → slow + fragile. A2 – Report: End Sep → VP review presenting quality gaps + mitigation plan. A3 – Redesign: Rebuilt 2nd workflow (Ports/Adapters; Lambda OOP; Step Fn for durability only). A4 – Deliver: Feb release on time + reusable testing/obs template for others. R: Dec delay acknowledged early (no surprise); Feb delivered on schedule, 0 stuck claims (~ $20 M saved). Lesson → lead with early risk escalation on all later projects. TRAITS: Execution • Ownership • Decision (Q 16 18 24)

---

Claims Modernization – Conflict Resolution Between Tech Leads

S: Jan 2025. Two tech leads debating Lambda granularity (blocking workflow release). T: Resolve conflict in 1 wk and standardize pattern across teams. A1 – Facilitate: Set joint session reviewing Step Fn→Lambda RPC semantics. A2 – Clarify: Compared approach to gRPC oneOf pattern; showed state cost of extra Lambdas. A3 – Decide: Consensus → single Lambda per activity type for traceability. A4 – Scale: Published design standard repo used by 5 teams. R: Conflict closed pre-launch; consistent Lambda pattern org-wide. TRAITS: Collaboration • Influence • Communication (Q 37 11 12)

Fraud Defense – Team Transformation

S: 2024 Q3 → Q1 2025 | Dist. Eng., Capital One Fraud Defense Authoring. Team (12 engs) migrating EMR → Glue without big-data skills; system unstable, 5000 line YAML pipelines. T: Within 3 months, stabilize Glue jobs and build a self-sufficient team able to ship new fraud defenses independently.

A0 – Assess: 1 wk audit → Spark UI unused; jobs ran 1 executor (qa config in prod). A1 – Upskill: Built hands-on Spark/Glue curriculum + weekly labs. A2 – Modernize: Introduced DBT for SQL defenses (modularity + testing + lineage). A3 – Simplify: Replaced custom StepFn orchestration with Glue Workflows (-60 % complexity). A4 – Mentor & Scale: Paired tech lead through 3 cycles → independent + coaching others.

R: Incidents 40 → 12 /mo (-70 %), runtime ↓ 35 %, $25 M fraud prevented. Same training kit rolled out to 2 more teams. TRAITS: Leadership • Efficiency • Mentorship (Q 19 20 21 22 23 38)

---

Fraud Defense – Glue Pipeline Incident Recovery

S: Oct 2024. Fraud Defense Spark pipeline failed 1 week after credit-card launch → ≈ $3 M fraud loss. T: Contain incident within 24 h and prevent reoccurrence org-wide.

A0 – Assemble Response: Formed SWAT workgroup (Data, Infra, Fraud) same day; defined blast radius + owners. A1 – Root Cause: Analyzed Spark UI → 30:1 partition skew causing OOM on 1 executor. A2 – Stabilize Fast: Repartitioned data + autoscale executors; service restored in 18 h. A3 – Institutional Fix: Wrote Spark tuning guide + post-mortem template; added skew alerting dashboard. A4 – Upskill: Led debug workshops for 3 teams → incidents self-diagnosed after.

R: Outage resolved < 24 h, 0 recurrence 12 mo; playbook adopted by 4 teams + Fraud Platform on-call. Lesson → treat incident as systemic learning moment. TRAITS: Reliability • Collaboration • Leadership (Q 9 27 29 19 38)

---

VP Trust – Influence Without Authority

S: Jan → Sep 2024 | Dist. Eng., Capital One Core Modernization. Peer VP (20 teams) resisted Temporal orchestration → risk of deadline slip. T: Win buy-in within 9 mo and move his org to Temporal without authority.

A1 – Understand: 1-on-1s with VP + mgrs → fear of delay if tech failed. A2 – Build Proof: POC Temporal workflow → deploy 2 wk → 3 d speedup. A3 – Influence Chain: Engaged trusted mgrs early + co-ownership of approach. A4 – Pilot & Scale: Low-risk team 2 wk trial → success → rollout to 20 teams.

R: VP became champion; Temporal standardized across 20 teams; deploy time ↓ 70 %. Used same method to influence 2 other VPs. TRAITS: Influence • Leadership • Stakeholder Mgmt (Q 11 12 13 24)

---

VP Trust – Temporal Adoption for Core Mod

S: Q4 2024 | Dist. Eng., Capital One Core Modernization. Org leaning on event-driven SQS/Lambda design for core account workflows. T: In < 2 mo shift executive direction to Temporal orchestration as enterprise pattern.

A1 – Assess: Mapped 15 async steps / 8 integrations → fragile state mgmt & months debug risk. A2 – POC: Built Temporal Saga for account booking; auditable + rollback in minutes. A3 – Address Consistency: Demoed strong-read API via Temporal signal + read-side materialization. A4 – Align Execs: Presented latency ↓ 70 %, failure ↓ 90 %; reframed as risk reduction not tech debate.

R: Adopted as enterprise standard across Core Mod; 20 + teams migrated; code complexity ↓ 40 %. Pattern carried to later modernizations. TRAITS: Strategy • Influence • Exec Alignment (Q 12 13 31 40)

Perfect — here’s Batch 3 of 4, covering the Pinterest Infra Gov stories rewritten in the dense, tactical senior-staff style. Each is tight (≈ 3 min delivery), timeline-anchored, and framed to highlight trust-building, quality, efficiency, and initiative, per Uber’s behavioral grid.

---

Pinterest – Temporal Evaluation

S: May → Jun 2023 | Staff Eng., Infra Governance team. Partner team building entitlement platform leaned toward SQS/Lambda choreography. Risk = distributed state mgmt nightmare + no observability. T: In 3 weeks evaluate alternatives and influence architecture choice without authority.

A1 – Analyze Options Mapped SQS/Lambda, Airflow, Temporal → benchmarked state visibility, latency, error recovery. Found queues = manual compensation, poor traceability.

A2 – POC Built Temporal workflow for budget reservation Saga; demo rollback & state query. Validated low latency (< 200 ms vs 3 min Airflow).

A3 – Align Team Presented to entitlement leads; acknowledged their SQS expertise, framed Temporal as safer for distributed tx. Let them run both approaches 1 week → Temporal wins observability metrics.

A4 – Document Decision Wrote trade-off doc (pros/cons per option); submitted to architecture review board.

R: Team adopted Temporal; 6 mo debug saved, failure handling standardized. Decision template reused for 2 tech evaluations. TRAITS: Strategy • Influence • Decision Making (Q 7 11 31 40)

---

Pinterest – Budget Planning (Trust During Onboarding)

S: Aug → Dec 2023 | Month 1 of onboarding. Infra Gov team owning budget-aware provisioning. 80 tables + 50 pipelines duplicated; no tests; contractors underperforming. T: Ship production system in 5 mo and earn cross-org trust (Finance, Platform, Data).

A0 – Research / Trust Map Met each org → Finance wanted auditability, Platform accuracy, Service owners simplicity. Noted no shared vision.

A1 – Design w/ Quality Front-Load Documented trade-offs (MVC vs Ports/Adapters) → chose Ports/Adapters for testability of junior team.

A2 – Pre-Exec Pilot POC GraphQL hierarchical queries + Cedar authz for audits. Used early feedback to lock design.

A3 – Build / Launch 6-stage approval workflow, audit changelog, real-time GraphQL subscriptions. Integrated Docker Compose in Bazel for local testing.

A4 – Operationalize Runbooks + monitoring; paired w/ junior engs to hand off ownership.

R: Prod launch on schedule; 0 incidents; pattern reused by 2 systems. Established cross-org trust in first quarter. TRAITS: Execution • Trust • Leadership (Q 4 11 23 24 36)

---

Pinterest – DBT Migration (Improving Productivity)

S: 2023 | Staff Eng. Infra Gov. 50 Python/SQL pipelines 90 % duplication → velocity collapse. T: In 6 mo, replace copy-paste jobs with reusable DBT macros.

A1 – Research: Cataloged 50 pipelines → grouped by pattern (cleanup, join, agg). A2 – Plan: 3-phase migration (macro → native → multi-tenant). A3 – Pilot: AWS platform subset 15 pipelines; validated lineage + CI/CD. A4 – Scale: Built macro library + auto validation on PRs; ran team training session.

R: Dev cycle weeks → days; velocity × 3; pattern adopted by adjacent teams. TRAITS: Execution • Efficiency (Q 10 19 20)

---

Pinterest – Cost Hierarchy (Stepping Up Beyond OKR)

S: Mar → Jun 2023. Org reorgs broke 80 cost pipelines; Finance spent weeks manual fixing. Not on my OKR. T: Solve structural problem in under 3 mo w/out ownership mandate.

A1 – Research: Mapped roll-up logic across 50 pipelines; dup everywhere. A2 – Design: Proposed cost_center_hierarchy table (parent/child + effective dates). A3 – Align: Met Finance lead; demoed simple org-change update. A4 – Implement: DBT dimension + traversal macros → replaced 5 pipelines initially.

R: Org change weeks → minutes; pattern reused 50 pipelines; concept replicated to 2 other hierarchies. TRAITS: Initiative • Influence (Q 11 31 32)

Credimi – COVID Decision Under Pressure

S: Mar → May 2020 | CTO at Credimi (digital lender, €2 B loans issued). COVID lockdown paralyzed banks; SMBs needed funds in days. T: Within 6 weeks, scale platform from 400 → 1 300 loans / mo while keeping credit risk controls intact.

A1 – Rapid Assessment: Met Risk & Ops; identified bottlenecks; segmented requests (auto-approve / manual / reject).

A2 – Rule Design: Created auto-decision engine: good history + < €50 k → auto-approve; high-risk sectors → manual.

A3 – Stakeholder Buy-in: Presented to 5 bank partners; aligned on daily risk reporting + escalation process.

A4 – Pilot → Scale: Launched with 1 bank; validated defaults; scaled to all 5 within 3 weeks.

R: Delivered liquidity to SMBs on time; 1 300 loans / mo (3 × baseline), 97 % satisfaction; became Generali’s emergency-funding platform. TRAITS: Execution • Decision • Urgency (Q 17 26 33)

---

Sapient Bioanalytics – Learning from Failure

S: 2018 → 2019 | Head of Software, $15 M Gates Foundation project. Funder wanted analytics features; grantees needed simple data tools. T: Deliver usable platform in 4 months balancing funder vs user needs.

A1 – Early Mistake: Focused solely on Foundation feedback → ignored end users.

A2 – Miss & Impact: Released MVP meeting funder criteria but grantees couldn’t use it.

A3 – Recovery: Re-engaged users via bi-weekly office hours + shadow sessions; logged pain points.

A4 – Redesign: Simplified flows + tool integration; ran joint UAT with grantees.

R: Platform adopted 3 mo later; 40 % features redesigned; stakeholder trust restored. Lesson → always involve end users when funder ≠ user. TRAITS: Learning • Stakeholder Mgmt (Q 8 35)

---

Snapshot Generator – Org-Level Impact

S: May → Jul 2024 | Dist. Eng., Capital One Fraud Modernization. 10 teams (~80 engs) blocked: CDC init took 30 days per topic. T: In 12 weeks, enable instant startup of real-time read side to unblock Dec launch.

A1 – Investigate: Found nightly exports not point-in-time; 30 d lag unavoidable in current design.

A2 – Prototype: Built Spark-based snapshot generator using event history to rebuild state quickly.

A3 – Validate: Single-topic POC completed in hours vs 30 d; verified by 3 team leads.

A4 – Scale & Adopt: Parameterized logic across 33 topics; added monitoring + runbooks; shared docs org-wide.

R: Init time 30 d → 10 min; 10 teams unblocked; technique reused in 3 modernization programs. TRAITS: Execution • Optimization • Reliability (Q 2 10 15 34)