Every data incident, rebuilt as a timed challenge.
Real production outages — cloud-bill blowups, broken pipelines, latency drift — recompiled from post-mortems into hands-on sims. Open any one to read the problem and the systems involved, then diagnose it on the clock against expert keys.
10 challenges · 10 live · one free sim every month
four phase
Poke Until It Dies
It's 09:00. The Airflow UI is a wall of yellow — ~600 tasks queued across four teams, almost nothing running. Workers show single-digit CPU. The scheduler heartbeat is healthy. Nobody deployed anything. It looks like the whole platform is broken and idle at the same time, and three teams are already in your DMs asking why their morning loads haven't started.
The Anonymization That Wasn't
We published a hospital discharge dataset for research, anonymized to k=5 - every record indistinguishable from at least four others on age, ZIP and admission month. A health journalist just emailed our privacy office: they named a specific patient and their diagnosis, using only the public file and public information. The k-anonymity math checks out. The re-identification still happened.
The Regex Blind Spot
Our nightly Presidio scan certifies silver.support_tickets as PII-free, and the coverage dashboard has been solid green for weeks. This morning a data-science intern pasted a query result into Slack: full 16-digit card numbers and SSNs sitting in plain text in the notes column. The scan says zero. The data says otherwise.
The Ghost in the Snapshot
An external auditor re-ran a right-to-erasure verification for a user you deleted six weeks ago. Their email and home address came back. The live table is clean, the job logged SUCCESS, and the same pipeline has 'erased' 412 people this quarter.
The Hot-Shard Spiral
It's 14:03 on Black Friday. MegaToys' flash sale went live at 13:30. Checkout is green, payments are clearing — but the warehouse queue is empty. orders_fact is 41 minutes stale and the gap is growing. At 13:00 the pipeline was sub-30s fresh. Fulfillment, fraud, and the live 'low stock' banners all read this table. Sale ends 15:00. You're on call.
The $516k Gap
Friday, 08:47. Finance Slack just pinged: the Q2 revenue tile on the CFO Looker dashboard reads $4.21M, but the ERP reconciliation says $3.69M. Six days of refunds appear to have vanished somewhere between Postgres and the boardroom. The CEO walks into the board room at 09:00.
The 2 AM That Wasn't 2 AM
It's 08:10. The exec revenue dashboard opened board prep at 4.1× yesterday's real number. No task went red — the Airflow UI is a wall of green. But daily_revenue shows 26 runs that all fired between 01:58 and 03:40 last night, and fct_revenue has ballooned. Nobody triggered a backfill. Finance wants the real number before the 09:00 standup.
The Reversible Hash
We share a user_hash column with an analytics partner so they can join on users without seeing identities. This morning the partner sent back a spreadsheet mapping our user_hash values straight to real email addresses. We used SHA-256 - it's a one-way function. So how did they reverse it?
The Silver Shortfall
Thursday, 09:48. A finance analyst reconciling the daily GMV report against the payment processor finds Gold sitting $1.29M below settled revenue. The weekly exec revenue review starts at 10:00.
The Small-Files Tax
It's 06:47. Finance Slack pinged overnight: analytics GCS bill jumped 4× last week, breaking the quarterly budget. No outages, no paging. Dashboards still render. Board prep starts at 08:00. You're on call.
Ready to take the pager?
One free simulation every month, no card. Replays last forever.