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

Discord Engineering Manager Interview Questions

30 real practice questions for the senior-level Engineering Manager role at Discord (Communications / Consumer), 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 Discord 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
10 easy · 10 medium · 10 hard
Avg. answer time
~4 min

Behavioral Questions (6)

  1. 1.Tell me about a time your team hit a latency or reliability wall in a real-time system. How did you diagnose it, and what did you change?

    easy~3 min

    What interviewers look for

    • Candidate can articulate a concrete latency budget or SLA that was being violated — not just 'it was slow' but specific p99 numbers, tail latency behavior, or dropped messages.
    • Demonstrates systems-level debugging: tracing the problem from user symptom back to a root cause (e.g., head-of-line blocking, GC pauses, back-pressure misconfiguration) rather than treating it as a black box.
    • Shows awareness of the tradeoffs made — what was sacrificed for lower latency (e.g., consistency, durability) and whether those tradeoffs were communicated to stakeholders.
    • Reflects on what observability tooling or instrumentation they added so the team could catch regressions earlier — showing a 'leave it better than you found it' mindset.

    Likely follow-ups

    • What did your monitoring or alerting look like before you found the issue — and how did it change afterward?
    • If Discord's voice infrastructure had a similar issue affecting thousands of active voice channels, how would your debugging approach change at that scale?
    • Who else did you loop in, and at what point? Were there any calls that in hindsight you made too late?

    Company context

    Discord's Real-Time Systems Mastery principle sits at the core of what the company ships — the voice, video, and text platform serves millions of concurrent users with strict latency expectations. An EM at Discord doesn't just manage engineers who work on these systems; they need enough technical depth to guide triage, push back on architectural shortcuts, and understand when a latency regression is existential versus acceptable. This question is a baseline signal for that depth.

  2. 2.Walk me through a project where your team shipped something that spanned two or more different languages or runtimes. What made the cross-language boundary hard, and how did you manage it?

    easy~3 min

    What interviewers look for

    • Candidate describes a real interface boundary — e.g., an FFI, a message protocol between services in different runtimes, or a NIF — and can explain what made it technically challenging (type safety, serialization, error propagation).
    • Shows awareness of how different runtimes have different failure modes and how they managed that in the team's workflow — e.g., different testing strategies, observability gaps, or deployment complexity introduced by a polyglot stack.
    • Demonstrates they thought about team structure and ownership: who owned which layer, how knowledge was shared across language boundaries, and whether they actively rotated engineers across the boundary or siloed by language.
    • Mentions concrete tactics for reducing friction — shared schema definitions, contract testing, documentation of cross-language invariants, or pair programming across the boundary.

    Likely follow-ups

    • If you were onboarding a Python engineer onto a feature that required touching an Elixir service, what would your ramp plan look like?
    • What broke first at the language boundary, and what did that failure teach you about how you'd spec the interface next time?
    • How did you handle code review when reviewers might be expert in one language but not the other?

    Company context

    Discord's stack is intentionally polyglot: an Elixir messaging core running on the BEAM, Rust NIFs via Rustler for performance-critical paths, a Python API monolith, and TypeScript on the frontend. Engineers at Discord are expected to cross language boundaries when shipping features — a Python developer may need to contribute to an Elixir service. EMs must be able to lead teams that work this way and must anticipate the coordination costs, knowledge gaps, and interface brittleness that polyglot systems create. This question tests whether the candidate has managed that complexity before or will be learning it on the job.

  3. 3.Tell me about the largest concurrency or scale challenge you've led a team through. What broke, how did you find it, and what changed architecturally?

    medium~4 min

    What interviewers look for

    • Candidate describes a genuine concurrency failure mode — race conditions, thundering herd, lock contention, BEAM scheduler saturation, connection pool exhaustion — with enough technical specificity to distinguish it from a generic 'we needed to scale' story.
    • Demonstrates how they structured the team's response: who owned the investigation, how they triaged under production pressure, and how they communicated status to leadership and stakeholders during the incident.
    • Describes an architectural change that addressed root cause — not just a horizontal scale-out, but a structural change to how the system handled concurrency (e.g., actor model refactor, sharding strategy, back-pressure introduction, queue decoupling).
    • Shows they built durable team capability from the incident — load testing practices, chaos engineering, or concurrency-aware design reviews added afterward.

    Likely follow-ups

    • Discord's guild server model means a single 'server' can have hundreds of thousands of concurrent members. If you were scaling the membership fanout system, what concurrency primitive would you reach for first?
    • How did you know the architectural fix was actually correct and not just a patch that deferred the problem?
    • What would you tell a new EM joining Discord about how to build a team that reasons well about concurrency from the start?

    Company context

    Discord regularly handles millions of concurrent WebSocket connections — a single large guild server can have hundreds of thousands of concurrent members generating real-time events. The BEAM's actor model is central to how Discord's Elixir services handle this concurrency, and the engineering team must reason about BEAM scheduling, distributed coordination, and back-pressure. EMs at Discord need to have personally led teams through concurrency-at-scale challenges — not just delegated — because technical decisions at this layer have product-wide consequences. This question differentiates EMs with real systems experience from those who managed roadmap but stayed above the technical fray.

  4. 4.Describe a time you pushed your team to refactor something specifically for testability. What was the state of the code before, what did you change, and did it actually improve confidence in the system?

    medium~4 min

  5. 5.Walk me through the last time you hit a wall debugging something — something where the obvious approaches failed and you had to go deep into docs, source code, or community resources to get unstuck. What did you find, and how did it change how you work?

    hard~5 min

  6. 6.Tell me about a time a real-time system you owned degraded in a way that users felt but your monitoring didn't catch. How did you find out, how did you respond, and what did you change so it wouldn't happen again?

    hard~5 min

Technical Questions (6)

  1. 7.You're onboarding an engineer who's strong in Python but has never touched Elixir. They're being assigned to a service running on the BEAM. How do you set them up to be productive without slowing down the team?

    easy~3 min

  2. 8.A product team wants to add a feature to Discord servers that would require reading a user's full message history across all channels they've ever joined. Given our Cassandra data model, how would you push back on or re-shape that requirement?

    easy~4 min

  3. 9.Your team owns Discord's WebSocket gateway. A deploy goes out and connection drop rates spike 3x, but CPU, memory, and error logs all look clean. What's your investigation strategy?

    medium~4 min

  4. 10.You need to ship a new Discord Activities feature that embeds a third-party game inside a voice channel. The game client communicates over WebSockets and expects sub-100ms round-trip times. How do you architect the communication layer and what risks do you bring to your team's attention before starting?

    medium~5 min

  5. 11.Discord's message fan-out — delivering a message sent in a large server to tens of thousands of online members — is one of the hardest scaling problems we have. If you were redesigning the fan-out pipeline from scratch, what would your architecture look like and what would you be most worried about?

    hard~5 min

  6. 12.We're considering migrating a Python service that handles bot API request routing to Rust. As the EM, how do you evaluate whether to do it, how do you plan the migration, and what would make you stop?

    hard~5 min

System Design Questions (6)

  1. 13.How would you design the presence system that powers the green/yellow/idle dots you see on every Discord user? Assume tens of millions of concurrent users and presence needing to update across potentially thousands of shared servers.

    easy~3 min

  2. 14.Walk me through how you'd design the permission system for Discord servers — roles, channel overrides, category inheritance — so it stays fast at read time even as a server grows to hundreds of roles and thousands of channels.

    easy~3 min

  3. 15.Design the rate limiting system for Discord's bot API. Bots send millions of requests per minute, and a single popular bot can be deployed across hundreds of thousands of servers simultaneously.

    medium~4 min

  4. 16.Design Discord's voice channel infrastructure — specifically the system that decides which media server a user connects to when they join a voice channel, and keeps that routing working during server failures and traffic spikes.

    medium~4 min

  5. 17.Design the read-receipt and typing indicator system for Discord DMs at scale. These features need to feel instant but Discord has hundreds of millions of users sending direct messages daily.

    hard~5 min

  6. 18.Design a system that lets Discord detect and enforce that a user has been banned from a server even if they create a new account. This needs to work at Discord's scale without blocking legitimate new account creation.

    hard~5 min

Leadership Questions (6)

  1. 19.Tell me about a time you had to make a hard call about what your team would NOT build. How did you decide, and what happened to the thing you cut?

    easy~3 min

  2. 20.Describe a time you had to give a senior engineer feedback they didn't want to hear. How did you approach it, and what came out of the conversation?

    easy~3 min

  3. 21.Tell me about a time you had to align two teams that had genuinely competing technical opinions about how to build a shared system. How did you get to a decision without having authority over both teams?

    medium~4 min

  4. 22.Tell me about a time you had to set technical direction for your team in a domain where you were learning alongside them. How did you establish credibility without being the expert in the room?

    medium~4 min

  5. 23.You're a few weeks into managing a team and you realize the team has been shipping fast but has almost no test coverage on the parts of the system that matter most. The team is proud of their velocity. How do you change that without breaking what's working?

    hard~5 min

  6. 24.Your team owns a core Discord service and a post-mortem reveals the incident was made significantly worse because your on-call engineer didn't have enough context to diagnose it alone. The team is distributed across three time zones. How do you fix the structural problem, not just the immediate gap?

    hard~5 min

Problem Solving Questions (6)

  1. 25.Rough estimate: how many concurrent WebSocket connections is Discord handling globally right now? Walk me through how you'd get to a number.

    easy~3 min

  2. 26.Server Boosting is one of Discord's primary revenue drivers. If you saw Nitro subscription churn spike 15% in a single week with no product changes, where would you start your investigation?

    easy~4 min

  3. 27.Discord is considering adding end-to-end encryption to DMs. Walk me through how you'd estimate the engineering cost — in team-months — and what you'd want to understand before committing to a roadmap.

    medium~5 min

  4. 28.Discord Activities — embedded games in voice channels — need to feel instant when they launch. How would you measure whether launch latency is actually good, and what's your target?

    medium~5 min

  5. 29.If Discord wanted to reduce infrastructure costs by 20% without degrading user experience, where would you look first and how would you validate you'd actually found real savings?

    hard~5 min

  6. 30.Discord has hundreds of millions of registered users but a much smaller daily active base. How would you think about whether that gap is a product problem, a market reality, or something else entirely?

    hard~5 min

More Discord interview questions