BitGo Software Engineer Interview Guide

The 60-second brief

As a Software Engineer at BitGo, you are at the forefront of securing the world’s digital assets. You will build and maintain the critical infrastructure that enables institutional-grade security for cryptocurrency custody, trading, and settlement. Your work directly impacts how global financial institutions interact with blockchain technology, requiring a high degree of technical precision and architectural rigor.

If the seat owns a service boundary, scope your preparation toward failure behaviour rather than topology. Retrying over an at-least-once channel produces duplicates by construction, so a retry policy is only as safe as the idempotency key underneath it.

Focus areas

• Evolve APIs without breaking pinned SDK clients
• Scope every query and cache key by tenant
• Build at-least-once pipelines with explicit deduplication horizons

Overview

This role is not merely about writing code; it is about solving complex distributed systems problems and ensuring the reliability of financial transactions at scale. You will contribute to projects ranging from high-throughput API services and low-level blockchain integrations to the design of secure, scalable storage systems. You will work alongside engineers who prioritize security, performance, and modularity in an industry that never sleeps.

Where to focus your preparation

Treat the sections below as preparation areas and confirm the format with your recruiter.

Preparation focus

No round sequence has been reported for this company, so work the categories below and confirm the format with your recruiter.

What to demonstrate

• Breadth across SQL, experimentation and product reasoning
• Ability to state assumptions before choosing a method

How to prepare

• Drill the practice exercises below and time yourself
• Prepare three quantified stories about decisions you drove

Where candidates lose points

Treating a timed-out write as a failed write

A timeout says the response did not arrive, not that the work did not happen; the server may well have committed and then lost the connection. Retrying a non-idempotent create after a timeout is the standard way to end up with two of something, and those duplicates land precisely when the system is already degraded and least able to absorb them. The discipline is to treat a timeout as unknown: either the write carries an idempotency key so the retry is safe by construction, or the client re-reads authoritative state before deciding what to do, and the interface says unknown rather than showing a failure that invites a second click.

Serialising a tenant's writes through select ... for update on a single counter row

It is the first change that makes a counter correct, and it caps that tenant's write throughput at roughly one divided by the lock hold time. A transaction that takes the lock, makes a network call and then commits holds it for the entire round trip: at 2 ms that is about 500 writes per second for the whole tenant, and the largest tenants are exactly the ones that exceed it. The damage then spreads, because every waiter holds a database connection while it queues, so one hot tenant drains the shared pool and the symptom presents as a site-wide latency incident rather than as a lock problem. The repairs are to shrink the critical section to a single statement, to shard the counter into per-(tenant, hour) or per-(tenant, bucket) rows and sum on read, or to batch in memory and flush periodically while accepting the bounded loss that batching implies.

A queue or buffer with no bound

Every producer-consumer boundary needs a capacity and a policy for reaching it: block the producer, shed load, or drop the oldest entry. Unbounded buffering converts a temporary slowdown into memory exhaustion and hides the backpressure signal that would have revealed the consumer was falling behind.

Not asking what the system looks like if it dies halfway through

For any multi-step write, say what state remains if the process stops between step two and step three, and what brings it back: a single transaction, a saga with compensating actions, an outbox, or a reconciliation job. Partial failure is routine at any real call volume, so 'that shouldn't happen' is an answer with nothing behind it.

Questions & practice

Design a solution for a bus route problem using BFS.

Medium · data structures and algorithms

Approach

1. Restate the input: its shape, its size, and what is guaranteed about it.
2. Walk one small example through your approach before writing the whole thing.
3. State the target complexity and say which constraint rules the naive version out.

Follow-up

• Which test case would catch an off-by-one here?
• How does this change if the input no longer fits in memory?

Solve a variation of the TwoSum array problem.

Medium · data structures and algorithms

Approach

1. Name the brute-force solution and its complexity before improving on it.
2. Choose the data structure from the access pattern, not from familiarity.
3. Walk one small example through your approach before writing the whole thing.

Follow-up

• Which test case would catch an off-by-one here?
• What is the worst case, and how likely is it on real data?

Implement a solution for finding the shortest path in a graph.

Medium · data structures and algorithms

Approach

1. State the target complexity and say which constraint rules the naive version out.
2. Restate the input: its shape, its size, and what is guaranteed about it.
3. Name the brute-force solution and its complexity before improving on it.

Follow-up

• How does this change if the input no longer fits in memory?
• Which test case would catch an off-by-one here?

Given an array of strings, find all anagrams using a hashmap.

Medium · data structures and algorithms

Approach

1. Restate the input: its shape, its size, and what is guaranteed about it.
2. Name the brute-force solution and its complexity before improving on it.
3. Choose the data structure from the access pattern, not from familiarity.

Follow-up

• What is the worst case, and how likely is it on real data?
• How does this change if the input no longer fits in memory?

Fold a deduplicated usage stream into hourly rollups

Easy · aggregation · deduplication · watermarks · exact-arithmetic

You are given one day of usage_event rows, up to 250 million, each carrying event_id, tenant_id, workspace_id, environment, sku, quantity numeric(20,6), idempotency_key, occurred_at and ingested_at. Produce usage_rollup_hourly cells keyed (tenant_id, workspace_id, sku, hour_start) with quantity_sum, event_count and source_max_ingested_at. An event counts once per (tenant_id, idempotency_key). The rollup grain has no environment column, so state your filter. One pass. Give your time and space bounds, and say what the deduplication actually costs in memory.

Approach

1. Bucket on occurred_at, never ingested_at: hour_start = date_trunc('hour', occurred_at at time zone 'UTC'). The two columns answer different questions. occurred_at says which hour the customer is billed for; ingested_at says how current the fold is. Using the second for the first makes late data invisible instead of correctable.
2. The fold is trivial and the deduplication is the entire cost, so price it before designing anything clever. An exact set over (tenant_id, idempotency_key) at 250M entries, stored as a 16-byte 128-bit hash in an open-addressed table at 0.7 load factor, needs about 357M slots at 16 bytes each, roughly 5.7 GB. The fix is partitioning by hash(tenant_id) % P so each shard holds 1/P of the set and no tenant's keys straddle shards.
3. Rule out a Bloom filter as a replacement, in the right direction: a false positive reports 'already seen' for an event never seen, so you drop a real event and lose revenue with no error raised. It is usable only as a negative pre-filter in front of the exact set, where a miss is conclusive and a hit must fall through to the real lookup.
4. Accumulate in scaled integers, not binary floating point. numeric(20,6) admits values below 10^14, so one event scaled to micro-units can reach 10^20, past int64's 9.22 x 10^18; use a 128-bit or arbitrary-precision accumulator unless you first bound the per-event maximum. binary64 represents integers exactly only to 2^53, about 9.01 x 10^15, and cannot represent 0.1 at all, so two runs that sum in different orders disagree.
5. Carry source_max_ingested_at = max(ingested_at) over the events folded into each cell, and count event_count over accepted, post-dedup events. Without that watermark there is no way to prove later what a number did and did not include, which is the first question any reconciliation asks.
6. State the environment filter explicitly, because the rollup grain cannot record it. A fold that quietly includes staging bills non-production traffic; one that quietly excludes it loses a cost signal. Production-only is the billing answer, and either way it belongs in the job name and the output metadata. Complexity: O(n) time, O(distinct dedup keys) space, dominated by the dedup set rather than by the cells.

Worked solution (25 min)

1. Write both key tuples down before any code: dedup key (tenant_id, idempotency_key), cell key (tenant_id, workspace_id, sku, hour_start), with hour_start derived from occurred_at in UTC.
2. Build a 10,000-row fixture containing one event duplicated three times under the same idempotency_key, two events sharing an idempotency_key across different tenant_id values, one event whose occurred_at is two hours before its ingested_at, and one staging event inside an otherwise production cell.
3. Fold it and assert each of those four expectations separately rather than eyeballing a grand total.
4. Re-run with the input shuffled and diff the output files.
5. Size the dedup set for 250M keys using the load-factor arithmetic and write the number down next to the fixture.

Expected result: The triplicate contributes one event and its quantity once. The two same-key, different-tenant events both count, because the dedup key is the pair. The late event lands in the hour of its occurred_at while that cell's source_max_ingested_at advances to the later timestamp. The staging event is included or excluded per the stated filter and never silently.

Verify it yourself

• Shuffled input produces byte-identical cells, including event_count and source_max_ingested_at.
• The sum of quantity_sum across all cells equals the sum of quantity over the distinct dedup keys, computed independently.
• Swapping the accumulator to binary64 makes a fixture of 0.1-valued quantities disagree with the exact sum. That is a failing test, not a rounding note.

Follow-up

• A producer retries at 23:59:59 and the retry lands at 00:00:01. The unique index on the daily-partitioned table must include the partition key. What gets double-counted, and what is the smallest change that fixes it?
• The consumer acknowledges its batch before committing the fold. Which failure loses revenue now, and which arrangement duplicates instead?
• What makes a re-run over the same day produce byte-identical rollups?

Rebuild an hourly rollup with deduplication and late-arrival accounting

Hard · window functions · deduplication · event time vs ingest time · watermarks

From usage_event (event_id, tenant_id, workspace_id, environment, sku, quantity numeric(20,6), idempotency_key, occurred_at, ingested_at), produce the values usage_rollup_hourly should hold for one tenant over one day: per (workspace_id, sku, hour_start) the deduplicated quantity_sum, event_count and source_max_ingested_at, bucketed by occurred_at. Duplicates share (tenant_id, idempotency_key). Also report, per hour, the running total across the day and the share of quantity that arrived more than two hours after the hour began. Write the query, and state which duplicates a daily unique index cannot catch.

Approach

1. Deduplicate in its own CTE before any aggregation, because a SUM cannot be un-summed: row_number() over (partition by tenant_id, idempotency_key order by ingested_at, event_id) = 1. Include the tiebreaker. Without it the surviving row is non-deterministic when two duplicates share an ingested_at, and a rollup described as deterministically recomputable then disagrees with itself between runs.
2. Bucket on occurred_at and nothing else, and pin the timezone explicitly. date_trunc('hour', timestamptz) truncates in the session's TimeZone setting, so the same query run by a session set to a non-UTC zone buckets differently; use the three-argument date_trunc('hour', occurred_at, 'UTC') on PostgreSQL 16 or later, or date_trunc('hour', occurred_at at time zone 'UTC') at time zone 'UTC' before that. Filter environment = 'production' explicitly, since metering covers three environments and billing covers one.
3. Aggregate to the grain with sum(quantity), count(*) and max(ingested_at). The last is not decoration: it is the watermark the row consumed up to, and without it there is no way to prove afterwards what a number did and did not include.
4. Compute the late share inside the dedup-and-aggregate step as a conditional aggregate, sum(quantity) filter (where ingested_at > hour_start + interval '2 hours'), then divide by the hour's total. Compute the running total as a window over the already aggregated rows: sum(quantity_sum) over (partition by workspace_id, sku order by hour_start rows between unbounded preceding and current row). Running either over raw rows puts the duplicates back.
5. Answer the index question exactly. The unique constraint is on (ingested_day, tenant_id, idempotency_key), because a unique index on a partitioned table must contain the partition key. It therefore deduplicates only within one ingest day and admits a duplicate whose retry crosses midnight or whose replay runs a week later. That is why this CTE dedups across the whole window being recomputed, and why the dedup horizon is a correctness parameter rather than a retention cost.
6. Keep the numeric type all the way through. quantity is numeric so the sums are exact; a cast to double precision anywhere in this pipeline reintroduces drift that surfaces only as a few unreconcilable cents per tenant per month, long after the query is out of anyone's mind.

Worked solution (40 min)

1. Generate a day of events for one tenant across three workspaces and two SKUs, inject 2% duplicates sharing an idempotency_key with a later ingested_at, and inject 1% whose ingested_at is six hours past occurred_at.
2. Write the dedup CTE alone and confirm its row count drops by exactly the injected duplicate count.
3. Aggregate and compare quantity_sum against a ground truth computed outside SQL from the generator's own records.
4. Add the two window and filtered aggregates, and confirm the final hour's running total equals the ungrouped day total for each (workspace, sku).
5. Run the whole query twice over static data and diff the two outputs.
6. Re-run with the session TimeZone set to a non-UTC zone and confirm hour_start does not move.

Expected result: One query: a dedup CTE using row_number() over (partition by tenant_id, idempotency_key order by ingested_at, event_id); aggregation by (workspace_id, sku, hour_start) producing quantity_sum, event_count and max(ingested_at) plus a filtered late-quantity aggregate; a running total as a window over the aggregated rows; and an explicit statement that the partition-scoped daily unique index cannot catch a duplicate that crosses the ingest-day boundary or arrives in a later replay.

Verify it yourself

• Row count after dedup equals raw count minus injected duplicates, exactly.
• The running total at the final hour equals the ungrouped day total for that (workspace, sku).
• Two consecutive runs over static data return byte-identical rows.
• Every hour_start is hour-aligned in UTC and unchanged by the session's TimeZone setting.

Follow-up

• A dispute forces the same recompute over 40 days for one tenant. What changes about the dedup CTE's memory use and the chosen plan, and what would you do about it?
• Two runs a minute apart return different quantity_sum values for an hour that is already closed. Give two mechanisms that produce that, and the single query that distinguishes them.
• Express the same rollup incrementally so it does not re-scan the day each time the watermark advances. What does the incremental version stop being able to answer?

Explain why the metering dashboard scans every daily partition

Medium · indexing · partition pruning · sargability · tenant skew

usage_event is range-partitioned daily on ingested_at and holds tenant_id, workspace_id, environment, sku, quantity numeric(20,6), occurred_at and ingested_at. The only relevant index is on (occurred_at). A dashboard runs select sku, sum(quantity) from usage_event where tenant_id = $1 and date_trunc('hour', occurred_at) >= $2 and environment = 'production' group by sku, and EXPLAIN shows a sequential scan of every partition. Give each distinct reason, rewrite the predicate so an index can serve it, propose the index, and state the write cost its column order adds.

Approach

1. Separate the three causes rather than blaming one. First, date_trunc('hour', occurred_at) wraps the column, so the predicate is not sargable against a btree on the bare column. Second, pruning keys off ingested_at while the query constrains occurred_at, so no partition can be excluded. Third, even made sargable, (occurred_at) is not tenant-leading, so for one tenant among thousands the scan reads the whole time range and discards nearly all of it.
2. Rewrite the bound carefully, because the obvious rewrite is only conditionally equivalent. date_trunc('hour', x) >= $2 equals x >= $2 only when $2 is already hour-aligned; for an arbitrary $2 it means x >= date_trunc('hour', $2) + interval '1 hour'. Normalise the parameter in the caller and leave the column bare.
3. Restore pruning with a second, redundant predicate on the partition key: ingested_at >= $2 - interval '<late-data horizon>'. State both sides of it. It prunes to a handful of partitions, and it silently omits any event whose ingest lagged past that horizon, which is precisely what a producer replay produces. Either document the horizon as a stated bound, or partition on occurred_at and move the problem into the dedup window instead.
4. Propose (tenant_id, occurred_at) include (sku, quantity) per partition. A partial index where environment = 'production' mostly saves size rather than selectivity, since production dominates the three environments; take it if non-production is a meaningful share and skip it otherwise.
5. Price the write path honestly. At roughly 250M rows/day each extra index is another insert plus WAL per row, and a tenant-leading key scatters inserts across one hot leaf per active tenant instead of appending to a single rightmost leaf, so page dirtying and random I/O both rise. An INCLUDE payload widens every leaf entry and enlarges the index accordingly.
6. Add the index-only-scan caveat before someone reports it as a regression: on a freshly appended table the visibility map is not yet set for recent pages, so the INCLUDE columns still cost heap fetches until autovacuum has been through, and the newest hour is exactly the data the dashboard reads.

Follow-up

• CREATE INDEX CONCURRENTLY is not supported on a partitioned parent. Give the sequence that gets this index onto 400 existing partitions without blocking ingest.
• One tenant holds 200 times the median row count and the dashboard still times out for them with the index in place. What changes?
• Should this read hit usage_rollup_hourly instead? State what that costs in freshness and what the watermark lets you promise.

A resumable usage export that never skips a row

Easy · pagination · indexing · tenant scoping · visibility

Customers pull their own rows from usage_event through GET /v1/usage to reconcile against their own systems. The table is append-only and partitioned daily on ingested_at; a large tenant adds millions of rows a day while the export is being walked, and a client may pause for hours and resume. Specify the cursor, the index it requires, the tenant scoping, the ordering guarantee you can honestly offer, the per-page cost as the walk deepens, and what the client must do to avoid missing rows.

Approach

1. Rule out LIMIT/OFFSET on two independent grounds and say both, because fixing only one leaves the other. Correctness: rows inserted between page requests shift the window, so a walking client skips rows and repeats others, which for a reconciliation consumer is silent data loss rather than an error anyone sees. Cost: the database still produces and discards the skipped rows, so page N costs time proportional to N x page_size and a deep page degrades from milliseconds to seconds.
2. Use keyset pagination over a stable, unique, indexed ordering: WHERE tenant_id = $1 AND (ingested_at, event_id) > ($2, $3) ORDER BY ingested_at, event_id LIMIT $4, carrying the last row's pair as the cursor. The row-value comparison navigates a composite btree directly, so each page is O(log n + page_size) and stays constant as the walk deepens. The precondition is that the cursor columns never change value for a row, which ingested_at satisfies and updated_at would not.
3. Lead the index with tenant_id - (tenant_id, ingested_at, event_id) - which is simultaneously the correctness guard and the plan choice. An index on (ingested_at) alone forces a filter across every tenant's rows, and on a table where one tenant holds most of them that is fine only for that tenant and terrible for everyone else. Because ingested_at is also the partition key, a resumed cursor prunes to the partitions from the cursor forward.
4. Name the visibility hazard rather than assuming it away: in PostgreSQL now() is transaction start time, so a transaction that starts at T, inserts, and commits at T+8 s writes a row whose ingested_at is T but which becomes visible only at T+8 s. A walker that has already passed T never returns it. The gap equals the writer's longest transaction, so the mitigation is either a safety lag - serve only rows older than now() minus the longest permitted transaction - or a client that re-walks a trailing overlap window and deduplicates on event_id, which is stable and unique.
5. Offer the guarantee you can actually keep: ordering by (ingested_at, event_id) with no claim whatsoever about occurred_at order, and completeness only behind the safety lag or with the documented overlap-and-dedup obligation on the client. Saying this in the API reference is part of the design, because the client's reconciliation logic is what has to absorb it.

Worked solution (20 min)

1. Write the keyset query and the exact index it needs, then read the query plan and confirm there is no Sort node.
2. Insert rows concurrently while walking with a LIMIT/OFFSET pager and count the distinct rows returned against the rows that exist; repeat with the keyset pager and compare.
3. Construct the out-of-order commit case by hand: open a transaction, insert, hold it open while the walker passes that timestamp, then commit, and check whether the walker ever returns that row.
4. Choose the mitigation - safety lag or client overlap plus event_id dedup - and write down the number it depends on.

Expected result: Keyset pagination on (tenant_id, ingested_at, event_id) with a matching composite index, constant per-page cost and no sort; the guarantee limited to ingested_at ordering with nothing claimed about occurred_at; completeness delivered by a safety lag equal to the longest permitted write transaction, or by a documented client overlap window with deduplication on event_id.

Verify it yourself

• The plan for a page should be an index scan with no Sort node. A Sort means the index column order does not match the ORDER BY and the constant-cost claim is false.
• Under concurrent inserts the offset pager and the keyset pager return different sets. Count the difference rather than reasoning about it.
• Your safety lag must be at least the longest transaction the writing service permits. If you cannot name that number, the completeness guarantee is not enforceable and should not be documented.

Follow-up

• The customer wants to reconcile by occurred_at instead of ingested_at. What breaks, and what would you offer them in its place?
• One tenant starts a full-history export. How do you keep it from occupying every connection in the pool?
• A customer reports a missing row. What do you check first, and what would each answer tell you?

An idempotent create endpoint that returns a one-time secret

Easy · idempotency · race conditions · api design

POST /v1/api-keys inserts a tenant_api_key row (tenant_id, workspace_id, name, key_prefix, secret_hash, scopes, status, auth_version, created_at, expires_at) and returns the plaintext secret exactly once, since only its SHA-256 is stored. Write volume is tens per second and clients retry on timeout. Design the idempotency mechanism: what the key is scoped by, where the record lives, its retention, what happens when the same key arrives with a different body, what happens when a retry arrives while the first request is still in flight, and what a replay returns for the secret.

Approach

1. Scope the key by tenant, not globally: uniqueness is on (tenant_id, idempotency_key), or one tenant's key collides with another's and the second caller receives a stored response for a request it never sent. Store a fingerprint of the request alongside it - method, path and a hash of the canonicalised body - so a mismatch is detectable.
2. Let the unique constraint decide the race instead of application logic. In the same transaction as the credential insert, run INSERT INTO idempotency_record (tenant_id, key, request_fingerprint, status) VALUES (...) ON CONFLICT DO NOTHING RETURNING id; no returned row means this is a replay, and the existing record is then read. A select-then-insert here loses to itself under concurrency in exactly the way this endpoint is meant to prevent.
3. Write the three replay outcomes as a decision table rather than as prose: same fingerprint and completed returns the stored response; same fingerprint and still in flight returns 409 with Retry-After, without blocking and without executing; different fingerprint returns 422, because replaying a stored response for a mutated body would tell the caller a credential was created for parameters it never sent.
4. Handle the secret as the part that makes this endpoint different from an ordinary idempotent create. The plaintext cannot be regenerated from secret_hash, so either the stored response holds it - making the idempotency record a secret at rest whose retention is now the secret's exposure window - or a replay returns the key metadata without the secret and the documentation says a lost response is resolved by listing keys and revoking the orphan. The second is the safer default precisely because credentials are listable and revocable.
5. Set retention from the client's retry budget, not from a round number: the record must outlive the SDK's maximum total retry duration, so twenty-four hours is defensible if that budget is minutes. After expiry the key is reusable and a very late retry creates a second credential, which is acceptable here only because the object is listable and revocable, and would not be for an unlistable side effect. Expire with a scheduled delete on an index over created_at.

Follow-up

• Two requests with the same key arrive concurrently. Show the exact statements and say which one loses, and how it finds out.
• The client receives a timeout, retries, and gets 409 in-flight. What should the client library do next, and for how long?
• How does the design change if the created object is not listable - a one-off payout, say - so an orphan cannot be found afterwards?

Invoice detail latency triples after an ORM relationship refactor

Easy · n+1 query · orm · latency regression

An invoice detail endpoint returned in 40 ms at p99 last week. After a refactor replaced a hand-written join with ORM relationship access it returns in 1.4 s, and the regression grows with the number of invoice_line_item rows on the invoice. Database CPU rose, but no statement in the slow-query log exceeds 3 ms. You have request traces with per-span SQL, the ORM statement log, and a staging copy of the data. Produce an ordered diagnostic checklist, the measurement that confirms the cause before any code change, and the fix.

Approach

1. Count statements per request before reading any statement duration. A slow-query log hides this class by construction, because every individual query is fast and only their number is wrong; take one trace and count SQL spans.
2. Establish proportionality rather than asserting it: sample invoices with 5, 20, 60 and 200 line items and plot statements per request against line count. A straight line of slope 1 through an intercept of one or two identifies a lazy relationship load, and no index or cache would move that line.
3. Locate the emitting attribute access in the refactored code and check whether the same shape repeats one level deeper, for instance a tax or adjustment collection hanging off each line, which turns the cost quadratic.
4. Fix with a bounded statement count: either one join that fetches invoice and lines together, or two statements where the second is WHERE invoice_id = $1 AND tenant_id = $2. Keep tenant_id in the predicate so the read stays tenant-scoped even though invoice_id already implies it.
5. Choose between the two deliberately: the join duplicates the wide parent row across N children on the wire, the two-statement form avoids that for one extra round trip. Prefer the join for narrow parents and the split for wide ones.
6. Pin it with a per-request statement-count assertion in a test that varies line count, because a latency assertion passes on a small fixture and would not have caught this.

Follow-up

• The endpoint now also needs per-line tax rows. Show the shape that keeps statement count constant instead of reintroducing the same defect one level down.
• How does this change if a transaction-pooling proxy sits between the service and the database, so each statement may land on a different backend session?
• The same page paginates invoices with LIMIT and OFFSET. Why is that a second, independent defect, and what replaces it?

Suggested practice checklist

For someone fluent in a dynamic language who has shipped real work but has never had to say what the runtime is doing underneath. The week is built on measuring and deliberately breaking things, because the questions that expose this background are the ones where the interviewer asks why a second time.

Day 1 — Measure before reasoning

• Take a slow piece of your own code, write down in advance where you believe the time goes, then profile it and record how wrong the guess was. The cost is usually an allocation you did not notice or an accidental quadratic membership test.
• Replace one list membership test inside a loop with a set and measure at a thousand, ten thousand and a hundred thousand elements, confirming the shape of the curve rather than only that it got faster.
• Write down the three quantities you can now measure instead of assert: wall time, peak memory, and call count for the function you suspected.

Deliverable: A before-and-after profile of real code plus a written note on the size of the gap between the guess and the measurement.

Day 2 — References, copies, and the bugs they produce

• Write the function with a mutable default argument, call it three times, and explain the accumulating result: the default is evaluated once when the function is defined, so every call shares one object.
• Build a nested structure, take a shallow copy, mutate an inner element, and show that both views changed, because a shallow copy duplicates the container and not the elements. Then fix it with a deep copy and state the cost you just accepted.
• Write two functions, one mutating its argument in place and one rebinding the local name, and predict the caller's view of each before running it. That single distinction produces most of the bugs that pass their tests.

Deliverable: Three small programs whose output you predicted correctly before running, each with a one-line statement of the rule underneath.

Day 3 — Types, once, in a language that checks them

• Port one module you have already written, roughly a hundred lines, into a statically typed language, and record every place the compiler demanded an answer your original had left implicit: a value that can be absent, a numeric width, a case never handled.
• Write the same signature in both languages and state what the static one guarantees before the program runs and what it does not, since it will not save you from a wrong algorithm or an index out of range.
• Write the difference between an interface satisfied by declaration and one satisfied structurally, with one case each where the other approach would miss the mistake.

Deliverable: One module in two languages plus a list of the questions the type checker forced you to answer.

Day 4 — Concurrency, starting with what actually runs at the same time

• Run the same CPU-bound function across four threads and four processes and measure both. Under the default CPython build the threaded version will not speed up, because only one thread executes bytecode at a time; the process version will. Check which build you are on first, since free-threaded builds remove that lock and change the result.
• Then run a blocking I/O workload across four threads and measure it speeding up, because the interpreter releases that lock around blocking calls, which is why treating threads as useless is wrong as a general claim.
• Build the lost update: two threads each incrementing a shared counter a hundred thousand times, and show a final value below the expected sum, because an increment is a load, an add and a store and the thread can be suspended between them. Fix it with a lock and then measure what the lock costs.

Deliverable: Three measurements, threads against processes on CPU work, threads on I/O work, and a demonstrated lost update, each with the mechanism written underneath.

Day 5 — Debugging as a procedure rather than an instinct

• Work one real failure as a bisection: find a revision or an input size where it is good and one where it is bad, halve repeatedly, and state the two assumptions bisection needs, that the property changes exactly once across the range and that the test is reliable.
• Minimise one failing input to the smallest version that still fails, and record how many rounds it took.
• Keep a hypothesis log for one bug in three columns, what I believe, what would disprove it, what I observed, and stop yourself the first time you are about to change two things at once.

Deliverable: One bug worked to root cause with a written hypothesis log and a minimised reproducing input.

Day 6 — Tests that catch the bug you are about to write

• Implement an LRU cache with a capacity bound, then write the three test cases that would catch an off-by-one in eviction: insert exactly capacity items and assert nothing was evicted, insert one more and assert the least recently used key is the one gone, and read an old key just before that insert so the eviction victim changes.
• Add a property test comparing your implementation against a deliberately slow reference, an ordered list scanned linearly, over a few thousand random operation sequences, because a slow reference finds the cases you would not have thought to write.
• Write one numeric test that fails under exact equality and passes with a tolerance, and state why the tolerance has to be relative rather than absolute once the magnitudes grow.

Deliverable: An LRU implementation with three boundary tests, one property test against a slow reference, and one tolerance-based numeric test.

Day 7 — Debug something broken, out loud

• Have someone plant three defects in a two-hundred-line program, an off-by-one, a shared mutable state bug, and a wrong error-handling path, then find them while narrating, under a fixed rule: state the hypothesis before touching anything.
• Time each one and record which tool found it, reading, a printed value, a debugger, or a test, because the question asked in interviews is how you would find it rather than what it was.
• Write the sentence you will use when you do not yet know the cause, one that names the next measurement instead of offering a guess.

Deliverable: A recorded debugging session with time-to-find per defect and the method that found each.

The failure that was yours

Keep one story where the bad call was yours rather than a dependency's or a manager's. Name the check that would have caught it, whether you added that check afterwards, and whether it has fired since. Answers that route blame outward end the conversation early; answers that end in a guardrail someone still relies on tend to open it up.

Own the incident where invoices undercounted metered usage

Medium · incident response · at-least-once · billing correction · postmortem

A metering consumer acknowledged each batch before committing the fold into usage_rollup_hourly. A rolling deploy restarted consumers mid-batch for two hours; roughly 1.4M usage_event rows were acknowledged and never folded, and 61 invoices sealed against the resulting rollups before anyone noticed. Take the owner's role. Describe an incident of comparable blast radius you owned: how it surfaced, the query that sized the loss, what you stopped first, and how the money was corrected. Give a wall-clock timeline and one thing you got wrong while it was still live.

Approach

1. Open with the invariant that broke and the direction of the error, because they determine everything else: acknowledging before committing makes the consumer at-most-once, so this loses events rather than duplicating them, and loss raises no error anywhere. A listener who hears 'we lost revenue silently' knows immediately why detection took two hours.
2. Size it with a stated reconciliation rather than an adjective: sum(quantity) from usage_event grouped by (tenant_id, sku, hour of occurred_at) over the window, against usage_rollup_hourly.quantity_sum on the same keys, filtered to environment='production' because staging and sandbox are metered but not billed. Then bisect by hour and tenant until single cells explain the gap. Say how long that ran and whether a replica could serve it while the incident was live.
3. Separate mitigation from fix and say which came first. Mitigation is holding the sealing job, because a sealed row is frozen by design and every minute of sealing converts a recoverable rollup into an invoice correction. The fix is moving the acknowledgement after the commit, which re-introduces duplicates that the dedup check on (tenant_id, idempotency_key) must now absorb.
4. State the correction path in the domain's own terms: sealed periods are never edited, so each affected tenant gets an adjustment line on the next invoice with kind='adjustment' and voided_by_line_id pointing at the line it reverses, priced against the same rate tier and carrying the watermark it priced against. That is four separate numbers — tenants affected, minor units, the cycle the adjustment lands in, and when customers were told.
5. Close on one prevention control with its cost, not five: a per-hour reconciliation comparing raw sum to rollup sum that pages above a threshold. Name the threshold and the false-page rate you accepted, because a detector nobody will keep staffed is not prevention.
6. Name a mistake you made inside the response window — the wrong first hypothesis, a mitigation that made it worse — rather than a design mistake from six months earlier. That is the part candidates rehearse away and interviewers weight heavily.

Follow-up

• Your fix moves the acknowledgement after the commit. What breaks now, and what absorbs it?
• One undercharged tenant has since churned. Do you bill them, and who decides?
• How would you have caught this in ten minutes instead of two hours, and what would that detector cost you in pages per week?

Estimate a tenant-leading index migration you have never run

Hard · estimation · online migration · index builds · uncertainty

Someone needs a date. usage_event carries an index on (occurred_at) and needs (tenant_id, occurred_at); the largest tenant holds roughly a hundred times the median tenant's rows, the table is partitioned daily with years of retention, and you have never run a migration on a table this large. Give an estimate you would defend: how you decompose the work, the two or three numbers you would go and measure first, the range and confidence you state, and what you commit to when the person asking needs a single date today.

Approach

1. Refuse the bare number and then give one anyway, in the form that is actually useful: a range plus the measurement that collapses it. 'Four to eleven days; one afternoon building this index on a restored copy of the largest partition takes that to within a day' is an answer, while 'it depends' is not.
2. Decompose by failure mode rather than into equal chunks, because that is where estimates go wrong. On a partitioned parent you create the index ON ONLY the parent, build each partition's index with CREATE INDEX CONCURRENTLY, then ALTER INDEX ... ATTACH PARTITION, at which point the parent index becomes valid. CONCURRENTLY does not block writes but scans each partition twice, waits out older transactions, cannot run inside a transaction block, and on failure leaves an invalid index you must drop concurrently and retry.
3. Name the two unknowns that dominate and price them: build time on one restored partition of realistic size, and whether the planner actually chooses the new index for the skewed tenant, since selectivity for a tenant holding most of the rows is a different question from selectivity for the median tenant. Both are half-day measurements against a replica, and both are cheaper than being wrong by a week.
4. State the assumptions the range is conditional on, because that is what makes a slip a re-estimate instead of a credibility event: no partition above a stated row count, one concurrent build at a time so it does not compete with ingest for I/O, and an ingest backlog that can absorb the added write amplification while both indexes exist.
5. Budget the step nobody budgets: verification and the old index's removal. Dropping the old index is fast, but deciding it is safe to drop means confirming no plan still uses it, and that confirmation waits on real traffic across a full weekly cycle rather than on your patience.
6. Answer the single-date request honestly. Commit to a date for the first checkpoint — the measured build number from the replica — and to re-estimating on that date, and say plainly what you are not committing to yet. A date with a scheduled re-estimate is worth more to the asker than a confident wrong one, and you should say why in those words.

Follow-up

• The concurrent build fails half way through the largest partition. What is the state of the database and what do you do next?
• Your estimate slips by sixty percent. Which assumption broke, and at what point would you have known?
• The person asking needs the date for a customer commitment. Does your answer change?

Reverse a webhook ordering decision after measuring its cost

Medium · reversing decisions · head-of-line blocking · at-least-once · api contracts

You argued for strict per-subscription ordering in webhook-delivery, which means one in-flight attempt per subscription. It shipped. Three months later a single unresponsive endpoint holds one subscription's queue at a six-hour backlog, and two customers report events arriving out of order anyway once their own retries are counted. Describe a decision you reversed: what you originally optimised for, the measurement that changed your mind, what the reversal cost in engineering time and customer change, and how you told the people who had already built on the original guarantee.

Approach

1. State the original decision as a trade you made knowingly. Ordering across a network requires a single in-flight attempt per subscription, and its price is head-of-line blocking whenever one endpoint is slow. 'We priced it wrong' is a much stronger opening than 'we did not realise', and it is usually the true one.
2. Bring the measurement that flipped it, not the anecdote: backlog age at the ninety-ninth percentile per subscription, the share of subscriptions where one slow endpoint gated an otherwise healthy queue, and the delivery throughput lost to serialisation. A reversal justified by complaints is indistinguishable from a reversal justified by fatigue.
3. Name what you learned about the guarantee itself, which is the engineering content of this story. At-least-once delivery means a retried event already arrives after newer ones and the consumer already must be idempotent, so a guarantee the customer has to defend against anyway was never worth what it cost to provide.
4. Describe the migration, because reversing a published contract is the hard half and the part candidates skip. Parallel attempts behind a per-subscription flag, a monotonically increasing sequence number added to the envelope so order-sensitive consumers can sort or discard, documentation that states at-least-once and unordered in those words, and a deprecation measured in quarters because the client is a pinned SDK inside a build pipeline you cannot see or redeploy.
5. Give the cost in the two currencies that matter: engineer-weeks, and how many customers had to change code. Then say who you told before it shipped rather than in a changelog afterwards, and which large customer you left on the old behaviour and for how long.
6. Close with the signal you now weight differently, stated as something you would do earlier next time: measuring the blocking cost on the slowest decile of endpoints before committing to the guarantee, rather than after a customer noticed.

Follow-up

• A customer insists they need ordering. What do you offer them that is not global serialisation?
• How did you choose the deprecation window given that you cannot see or redeploy the clients?
• What would have to be true for you to reverse back?