What Candidates Reported
A 2024 OpenAI loop report describes designing a Slack MVP and then extending message delivery for scale and reliability. A separate September 2025 Software Engineer report identifies Slack as the system-design question. The overlap supports company-specific practice while leaving room for interviewer-specific scope.
Problem and Scope
Design workspaces, channels, direct messages, live delivery, durable history, offline catch-up, multi-device read state, presence, search, and retention. Clarify ordering, acknowledgement, authorization, channel-size distribution, regions, and the MVP boundary before choosing WebSocket, queue, or database technology.
Critical Design Decisions
The core is a durable per-conversation message path with stable IDs and explicit acknowledgement semantics. Connection gateways should remain replaceable. Fan-out must adapt to channel size, reconnect must detect gaps, and ephemeral presence must not contaminate durable history. Authorization and deletion apply to search, caches, archives, and queued delivery as well as the primary store.
Failure Scenarios to Practise
Pressure-test a lost send acknowledgement, gateway failure, slow mobile device, reconnect storm, 500,000-member channel, membership revocation during queued delivery, home-region failure, and deletion after indexing. Trace the user-visible outcome, authoritative state, and recovery signal for each.
How the Answer Is Evaluated
MockMe scores scope and messaging semantics (15%), message path and state (25%), scale and fan-out (25%), reliability and governance (20%), and communication and trade-offs (15%). Strong answers keep durable correctness separate from live transport and explain large-channel and regional behavior concretely.