Intervu is in beta — feedback welcome at support@intervu.io

Affirm Engineering Manager Interview Questions

30 real practice questions for the senior-level Engineering Manager role at Affirm (Fintech), spanning behavioral, technical, system design, leadership, and problem solving. Lead engineering teams, manage people and processes, and drive technical strategy. The first 3 questions below include what Affirm interviewers actually listen for, plus likely follow-ups.

Practice these questions with AI feedback
Questions
30
Categories
Behavioral (6), Technical (6), System Design (6), Leadership (6), Problem Solving (6)
Difficulty mix
9 easy · 10 medium · 11 hard
Avg. answer time
~4 min

Behavioral Questions (6)

  1. 1.Tell me about a system your team owned that was a black box to the rest of the org — nobody could explain what it actually did or why. How did you fix that?

    easy~3 min

    What interviewers look for

    • Candidate can name the concrete system and articulate exactly why it was opaque — not just 'it was complex' but specific behavioral unpredictability, missing docs, or logic only one person understood.
    • Took active ownership of the simplification, not just asked an IC to refactor. Drove a process: identified stakeholders, surfaced trade-offs clearly, and made the system's behavior auditable by non-engineers.
    • Connected the opacity to a real downstream risk — a customer impact, a compliance exposure, or a repeated on-call incident — demonstrating that hidden complexity is treated as a defect, not a technical curiosity.
    • After the change, could describe how they validated that stakeholders now understood the system, e.g., a post-simplification design review, documentation, or runbook that non-engineers could follow.

    Likely follow-ups

    • What did you have to sacrifice in terms of technical capability or performance to make it simpler? How did you decide that trade-off was worth it?
    • Who was the hardest stakeholder to convince that simplification mattered here, and how did you make the case?
    • If that system had been in Affirm's underwriting or payment settlement stack, what additional concerns would you have raised before simplifying it?

    Company context

    Affirm's 'No Fine Print' value isn't just a consumer brand promise — it's an internal engineering constraint. Affirm builds regulated financial products where hidden system complexity has direct customer and compliance consequences: a loan term calculated by an opaque rule engine, a fee suppressed by an undocumented flag, or a settlement flow nobody can explain to a regulator. The Honest Engineering principle explicitly treats hidden complexity as a defect. This question surfaces whether the EM candidate has internalized that standard and operationalized it with their team, not just paid lip service to clean code.

  2. 2.Walk me through the last time a production incident led you across multiple team boundaries before you found the root cause. What did you touch, and how did you coordinate?

    medium~4 min

    What interviewers look for

    • Candidate can trace the incident across at least two distinct service or team boundaries — not just two microservices they own, but genuinely crossing into another team's domain — and describe what signals pointed them each direction.
    • Demonstrates active coordination with peer teams during the incident: how they established shared war rooms, divided diagnostic responsibility, and avoided siloed debugging that would have extended MTTR.
    • Shows understanding of the end-to-end transaction flow in their domain — e.g., could reason about how a failure in one system (auth, underwriting, ledger, repayment) propagates downstream before manifesting as a visible symptom.
    • After resolution, drove a cross-team post-mortem or COE (Correction of Errors) document that identified systemic gaps in observability or ownership at the boundary — not just fixed the immediate bug.

    Likely follow-ups

    • Where did the observability break down — what signals were missing that would have cut your investigation time in half?
    • How did you handle a peer team that was defensive or slow to engage during the incident? What did you do when their help was critical path?
    • What changed in how your teams instrument or monitor at service boundaries after this incident?

    Company context

    Affirm runs a vertically integrated financial stack — underwriting, checkout, payments, servicing, and capital markets are all internally owned and deeply coupled. A single loan origination touches credit decisioning, fraud scoring, bank integrations, ledger writes, and consumer notification systems. The Vertical Ownership principle exists because Affirm deliberately does not outsource these concerns to separate vendors, which means EMs must understand — and help their teams understand — the full lifecycle, not just their slice. This question tests whether the candidate has the cross-system intuition to lead investigations that cross team lines, which is essential at Affirm's scale.

  3. 3.Tell me about the last time you disagreed with a decision made by someone above you in the org. What was the decision, how direct were you in pushing back, and how did it end?

    medium~4 min

    What interviewers look for

    • The disagreement was substantive — about a real product, technical, or people decision with meaningful consequences — not a minor process preference or resourcing quibble.
    • Was direct and specific in the pushback: named the concern explicitly, stated what they believed the right decision was, and did so in the room or in writing — not in a side channel or only to peers.
    • After disagreeing, either changed their position based on new information (and can explain what changed their mind) or committed fully after losing the argument — no passive resistance or sandbagging after the decision was made.
    • Shows awareness of how to calibrate directness at different levels of the org — direct with their skip-level looks different from direct with a peer — without becoming less honest or burying the concern.
    • Reflects on what they'd do differently — either in how they framed the disagreement or in when they chose to escalate vs. accept the outcome.

    Likely follow-ups

    • How did you decide when to push back one more time versus accept the decision and move forward? Where's that line for you?
    • Was there anyone else on your team or peer level who shared your concern but didn't say anything? How did you handle that dynamic?
    • Has there been a case where you pushed back, lost, and were proven right afterward? How did you handle being right in a way that didn't damage the relationship?

    Company context

    Affirm's Direct Communication principle isn't a cultural nicety — it's a deliberate counterweight to the political dynamics that grow in any scaling company. Affirm explicitly values engineers and managers who name trade-offs clearly, disagree in the open, and commit cleanly after losing. This is especially important for senior EMs who are expected to represent their team's technical and ethical concerns to leadership, including on topics like credit model fairness, consumer data use, or regulatory compliance, where the stakes of polite ambiguity are real. Affirm's interview process probes this because passively compliant managers are a liability in a company that runs on honest finance as its brand promise.

  4. 4.Tell me about a feature you shipped where the biggest risk wasn't technical — it was credit, fraud, or regulatory exposure. How did you catch it, and what did you do about it?

    hard~5 min

  5. 5.Describe a time when your team's system was causing downstream confusion — other teams couldn't predict its behavior or couldn't trust its output. How did you diagnose why the system was confusing, and what did you change?

    hard~5 min

  6. 6.Walk me through a time when a customer-facing bug turned out to have its root cause in a system your team didn't own. How far did you have to dig, and what did you do when you hit the boundary of your team's scope?

    hard~5 min

Technical Questions (6)

  1. 7.Your team owns a Kafka consumer that processes loan repayment events. A downstream team tells you their ledger balances are occasionally off by a few cents. How do you start debugging that?

    easy~3 min

  2. 8.You're onboarding a new engineer onto your team and they ask why your services use gRPC internally instead of REST. What's your answer, and where would you still choose REST?

    easy~3 min

  3. 9.We're adding a post-purchase BNPL feature to the Affirm Card — a user taps their card, the charge posts, and then within 24 hours they can elect to split it into installments. Walk me through the key architectural decisions you'd have to make to support that flow.

    medium~5 min

  4. 10.Your team's Python underwriting service is hitting CPU saturation at peak checkout volume. Before you reach for horizontal scaling, what would you instrument and investigate first?

    medium~4 min

  5. 11.You're designing the data pipeline that feeds real-time features into Affirm's underwriting model at checkout. A data scientist wants sub-100ms feature freshness on user payment history. Walk me through how you'd architect that.

    hard~5 min

  6. 12.You need to migrate a core loan servicing table in PostgreSQL — 200 million rows, actively written by multiple microservices — to add a new column required for an upcoming compliance deadline. How do you do this without downtime?

    hard~5 min

System Design Questions (6)

  1. 13.Walk me through how you'd design the repayment reminder notification system for Affirm Pay-Over-Time. Users have loans with varying due dates, and a missed payment is a real financial harm — how do you make sure the right message gets to the right person at the right time?

    easy~3 min

  2. 14.Design the merchant integration layer for Adaptive Checkout — the piece that sits between a merchant's checkout page and Affirm's underwriting engine. Merchants range from a single Shopify store to Walmart. How do you build it to serve both ends of that spectrum?

    easy~4 min

  3. 15.Design a real-time fraud scoring service that sits in the loan origination path for Affirm Pay-Over-Time. Every checkout triggers it, it can't add more than 50ms to the p99 latency, and a false positive means a real consumer gets denied credit. Walk me through your approach.

    medium~4 min

  4. 16.Design the ledger system that tracks balances and transactions for the Affirm Money Account. Users can hold a balance, earn yield, and spend directly via the Affirm Card. What are the hardest correctness problems in that system and how do you solve them?

    medium~5 min

  5. 17.Design a system that lets Affirm dynamically configure and personalize loan offer terms — interest rate, term length, down payment — for each checkout, across thousands of merchants simultaneously, without requiring a code deploy to change any parameter. How do you build that configuration plane?

    hard~5 min

  6. 18.Design the reconciliation system that ensures Affirm's internal loan ledger stays in sync with external banking rails — ACH, card networks, and partner bank settlement. At Affirm's scale, with millions of transactions a day, what does that pipeline look like and how do you detect and resolve discrepancies?

    hard~5 min

Leadership Questions (6)

  1. 19.Tell me about a time you had to staff up a team quickly — new headcount approved, roles to fill, and real delivery pressure. How did you decide who to hire and how did that affect team dynamics?

    easy~3 min

  2. 20.Describe a time you had to set a technical direction for your team that the team initially didn't agree with. How did you build alignment, and what did you do if they were right?

    easy~4 min

  3. 21.You're an EM at a company that processes millions of financial transactions a day and a post-mortem reveals your team's service caused real monetary harm to users — overcharges, missed credits, something tangible. How do you lead through that?

    medium~4 min

  4. 22.Tell me about a time you had to advocate for a significant engineering investment — refactoring, observability, reliability work — to a product or business stakeholder who didn't see the urgency. What was your argument and did it work?

    medium~4 min

  5. 23.You're managing a team at the intersection of two large initiatives that have competing resource needs — both PMs think your team is primarily theirs. How have you handled that kind of priority conflict, and what did you do when the tiebreaker wasn't clear?

    hard~5 min

  6. 24.Tell me about the hardest performance conversation you've had with an engineer — not a pip, but the conversation before you knew if it was going to become one. What made it hard, and what did you say?

    hard~5 min

Problem Solving Questions (6)

  1. 25.Estimate how many BNPL loan offers Affirm serves in a single Black Friday. Walk me through your assumptions and where you think the load concentrates.

    easy~3 min

  2. 26.Your team's authorization rate on Affirm Pay-Over-Time drops 4% over a weekend with no deploys and no incidents filed. How do you figure out what happened?

    easy~4 min

  3. 27.Estimate the annual cost to Affirm of a 10-basis-point increase in default rate across the Pay-Over-Time portfolio. What inputs do you need and how do you structure the math?

    medium~4 min

  4. 28.Affirm wants to expand Pay-Over-Time into a new country. What's your framework for estimating whether the engineering investment required is worth pursuing, and what Affirm-specific risks would you flag before committing your team?

    medium~5 min

  5. 29.How would you measure whether Adaptive Checkout's personalization is actually driving incremental loan origination, versus just capturing demand that would have converted anyway? Walk me through the experiment design and the metrics you'd trust.

    hard~5 min

  6. 30.Affirm is considering allowing users to increase their loan limit mid-repayment — essentially offering a credit line increase while an existing loan is active. Before your team writes a line of code, what analytical framework would you use to size the risk, and what's the first thing you'd want to know about the existing repayment data?

    hard~5 min

More Affirm interview questions