Rotating the Box (LeetCode 1861) — Simulate a box tilting to the right so that stones fall until they hit an obstacle or the floor; rearrange the grid to reflect the final state.

The key insight is to process each row right-to-left: maintain a pointer for the next open position and shift stones there when encountered, leaving obstacles in place, then rotate the whole matrix 90 degrees clockwise. Time O(m*n), space O(m*n) for the output. A likely follow-up is asking you to support left or upward tilting, testing whether your logic is direction-agnostic. Array/two-pointer mastery, in-place manipulation vs auxiliary space tradeoff, and handling multiple constraints (stones, obstacles, gravity) without conflating them.

Word Break (LeetCode 139) — Given a string and a dictionary of words, determine whether the string can be segmented into a space-separated sequence of one or more dictionary words.

Classic 1-D DP: dp[i] is true if s[0..i) can be segmented. For each position i, check all j < i where dp[j] is true and s[j..i) is in the word set (as a hash set for O(1) lookup). Overall O(n^2) time, O(n) space. Follow-up: return all valid segmentations (backtracking with memoization). DP formulation from first principles, ability to reason about substring boundary conditions, and clean handling of the base case dp[0]=true.

Squirrel Simulation (LeetCode 573) — Given a tree position, a squirrel starting position, and a list of nut positions, find the minimum total distance the squirrel must travel to collect all nuts and bring them to the tree.

The squirrel must eventually make a round-trip (nut to tree) for every nut, giving a baseline of 2 * sum(distances from tree to each nut). The squirrel gets one free deviation: for one nut it travels squirrel-to-nut instead of tree-to-nut, so the optimal choice maximizes (dist(tree, nut) - dist(squirrel, nut)). Time O(n). Follow-up: handle multiple squirrels or weighted paths. Greedy insight that only one nut deviates from the optimal base path, Manhattan distance computation, and ability to reformulate as a savings-maximization problem.

Design a hotel booking system (reported directly by interviewing.io as a previously common Block system design question; confirmed across multiple candidate summaries).

Start with requirements: search by date/location, hold/reserve/cancel a room, handle concurrent bookings. Key entities: Hotel, Room, Reservation. The central challenge is preventing double-booking: use optimistic locking or a database-level unique constraint on (room_id, date_range), or a distributed lock for the hold-then-confirm flow. Discuss CAP tradeoffs — availability vs consistency for reservation windows. Follow-up: how do you handle the hold expiring if the user abandons checkout? Structured requirements gathering, identification of the concurrency problem as the crux, and a concrete (not hand-wavy) mechanism for preventing double-booking.

Design an offline-capable Point of Sale system: how does a Square Register keep working when it loses internet connectivity, and how does it sync and resolve conflicts when connectivity returns?

Local-first architecture: all transactions written to local storage (SQLite or equivalent) with a pending queue. On reconnect, replay the queue against the server with idempotency keys. Conflict resolution: server is authoritative for card authorization; local cash transactions can be accepted optimistically. Discuss vector clocks or last-write-wins for metadata conflicts. Follow-up: how do you handle a card that was valid offline but gets declined on sync (stolen card reported during outage)? Understanding of local-first / offline-first patterns, idempotency in the sync protocol, and thoughtful handling of the fraud/auth edge case unique to payments.

Design a card-present payment-processing pipeline for Square handling ~10,000 TPS with a 100ms end-to-end authorization budget.

Trace the full flow: card reader -> gateway -> card network (Visa/MC) -> issuing bank -> back to merchant. Identify the latency bottlenecks (network round-trip to card network is the dominant cost). To hit 100ms: co-locate gateway with network endpoints, use connection pooling to card networks, implement fast-path for common issuers. For 10K TPS: horizontal stateless gateway tier, async logging, circuit breakers per issuer. Follow-up: how do you handle partial outages at a specific card network? Domain literacy around card-network rails and PCI boundaries, ability to budget latency across hops, and horizontal scaling reasoning rather than single-machine optimizations.

Tell me about a time you disagreed with a teammate's technical decision. How did you handle it and what was the outcome?

Use STAR. The strongest answers show that you raised the disagreement directly and early (not passively), backed your position with data or a concrete tradeoff, and were willing to commit to the team decision once made even if you disagreed. Block values direct communication and psychological safety, so avoid stories where you just went along silently. Candor and directness in raising disagreement, evidence of good-faith engagement with the other perspective, and ability to separate ego from outcome.

Tell me about a time you took ownership of a problem that wasn't assigned to you.

Pick a story where the problem had real business impact and fell into a gap. Show that you identified the gap proactively (not reactively), scoped it, fixed it, and communicated the resolution. Quantify impact where possible. Block's mission around economic empowerment means they want engineers who think beyond ticket scope. Intrinsic motivation vs assignment-following, judgment about when to escalate vs self-solve, and a concrete positive outcome.

Describe a production incident you owned end-to-end — from detection through root cause to prevention. What did you change afterward?

Structure: detection (how did you find out?), triage (how did you scope impact?), mitigation (what did you do first to stop the bleeding?), root cause (what actually caused it?), prevention (what structural change did you make?). The best answers distinguish mitigation from fix and include a blameless postmortem culture point. Systematic debugging process, ownership through the full lifecycle rather than just the fix, and proactive systemic improvement vs one-off patch.

Block Interview Questions

Q: Squirrel Simulation (LeetCode 573) — Given a tree position, a squirrel starting position, and a list of nut positions, find the minimum total distance the squirrel must travel to collect all nuts and bring them to the tree.

The squirrel must eventually make a round-trip (nut to tree) for every nut, giving a baseline of 2 * sum(distances from tree to each nut). The squirrel gets one free deviation: for one nut it travels squirrel-to-nut instead of tree-to-nut, so the optimal choice maximizes (dist(tree, nut) - dist(squirrel, nut)). Time O(n). Follow-up: handle multiple squirrels or weighted paths. Greedy insight that only one nut deviates from the optimal base path, Manhattan distance computation, and ability to reformulate as a savings-maximization problem.

Q: Design a hotel booking system (reported directly by interviewing.io as a previously common Block system design question; confirmed across multiple candidate summaries).

Start with requirements: search by date/location, hold/reserve/cancel a room, handle concurrent bookings. Key entities: Hotel, Room, Reservation. The central challenge is preventing double-booking: use optimistic locking or a database-level unique constraint on (room_id, date_range), or a distributed lock for the hold-then-confirm flow. Discuss CAP tradeoffs — availability vs consistency for reservation windows. Follow-up: how do you handle the hold expiring if the user abandons checkout? Structured requirements gathering, identification of the concurrency problem as the crux, and a concrete (not hand-wavy) mechanism for preventing double-booking.

Q: Design an offline-capable Point of Sale system: how does a Square Register keep working when it loses internet connectivity, and how does it sync and resolve conflicts when connectivity returns?

Local-first architecture: all transactions written to local storage (SQLite or equivalent) with a pending queue. On reconnect, replay the queue against the server with idempotency keys. Conflict resolution: server is authoritative for card authorization; local cash transactions can be accepted optimistically. Discuss vector clocks or last-write-wins for metadata conflicts. Follow-up: how do you handle a card that was valid offline but gets declined on sync (stolen card reported during outage)? Understanding of local-first / offline-first patterns, idempotency in the sync protocol, and thoughtful handling of the fraud/auth edge case unique to payments.

Q: Design a card-present payment-processing pipeline for Square handling ~10,000 TPS with a 100ms end-to-end authorization budget.

Trace the full flow: card reader -> gateway -> card network (Visa/MC) -> issuing bank -> back to merchant. Identify the latency bottlenecks (network round-trip to card network is the dominant cost). To hit 100ms: co-locate gateway with network endpoints, use connection pooling to card networks, implement fast-path for common issuers. For 10K TPS: horizontal stateless gateway tier, async logging, circuit breakers per issuer. Follow-up: how do you handle partial outages at a specific card network? Domain literacy around card-network rails and PCI boundaries, ability to budget latency across hops, and horizontal scaling reasoning rather than single-machine optimizations.

Block (formerly Square) runs a 5-7 round interview loop that emphasizes practical pair programming over whiteboard LeetCode, conducted in CoderPad. The loop typically includes a recruiter screen, a technical phone screen, two coding rounds, a system design round, a hiring manager round, and a behavioral/leadership round (the last two primarily for mid-to-senior levels). Difficulty is rated around 3.1/5 by candidates, and Block is widely noted for providing post-interview feedback. Since the company reorganized into multiple business units (Square, Cash App, Afterpay, Bitkey), system design and domain questions have become increasingly BU-specific around payments, fintech, and crypto.

18 real questions across 3 rounds

Reported via Taro (jointaro.com) Block SWE interview experiences (2024-2025), Glassdoor Block SWE interview summaries (search snippet aggregates, 2024-2025), interviewing.io Block/Square interview guide (2023), coditioning.com Square pair programming guide, Blind discussions on Square pair programming format.

Run a Block mock with the AI interviewer. It asks questions like these, follows up on your answers, and tells you exactly what to fix. The only score that means anything is the one on your real answers.

Continue with Google

Keep prepping

All company interviews