
Moved Pune to Bengaluru in June. The March order must still show Pune. Which design? A) UPDATE the address B) New row per move, with dates C) Join on is_current D) previous_address column Reply A-D. leetdata.ai/modeling/scd-add… #SQL

Weekly Data Engineering Newsletter. Subscribe to dataengineeringweekly.com | Wanna talk about Data engineering? Book Me here calendly.com/apackkildurai

Moved Pune to Bengaluru in June. The March order must still show Pune. Which design? A) UPDATE the address B) New row per move, with dates C) Join on is_current D) previous_address column Reply A-D. leetdata.ai/modeling/scd-add… #SQL

ROWS BETWEEN 6 PRECEDING counts 7 rows, not 7 days. Skip a day of sales and your "weekly" sum reaches back weeks. Use RANGE BETWEEN INTERVAL 6 DAY PRECEDING AND CURRENT ROW. leetdata.ai/problems/range-f… #SQL

The schema rule each app tests in a modeling round: Ride-hailing: 2 FKs to one locations table Payments: append-only events Hotel: no overlapping stays per room Social: UNIQUE (follower_id, following_id) leetdata.ai/modeling #DataEngineering

Should list customers who never ordered. Returns 0 rows. Why? WHERE id NOT IN (SELECT cust_id FROM orders) A) Subqueries not allowed B) NULLs in cust_id C) Needs DISTINCT D) Needs a cast Reply A-D. leetdata.ai/problems/not-in-… #SQL

Daily report says 1 order on a day nobody ordered. calendar LEFT JOIN orders, then COUNT(*). The empty day still has its calendar row, so COUNT(*) = 1. COUNT(o.order_id) gives 0. leetdata.ai/problems/calenda… #SQL

9 SQL topics, one problem each: casting, % of total, 2nd highest salary, semi join, top 2 per group, crosstab, 15-min buckets, sentinel NULLs, new vs returning users. The one that stalls you is this week's topic. leetdata.ai/problems #SQL

Same room, bookings a and b. Which WHERE catches every overlap? A) a_in BETWEEN b_in AND b_out B) a_in < b_out AND b_in < a_out C) a_in = b_in D) a_out > b_in Reply with a letter. leetdata.ai/problems/date-ra… #SQL

Data Engineering Weekly is on WhatsApp. You get the weekly issue, SQL interview problems in order, and system design questions from real company write-ups. A channel, not a group. Your number stays hidden. whatsapp.com/channel/0029Va9…

Two jobs append to one Iceberg table from snapshot 41. A commits first. B: A) overwrites A's files B) retries on the new snapshot, both land C) breaks it till compaction D) forks into 2 snapshots leetdata.ai/paths/lakehouse #dataengineering

8 join problems, in order: inner, LEFT + IS NULL, anti-join, NOT IN + NULL trap, full outer, multi-condition, date range, many-to-many. Number 4 returns zero rows, no error. Free, in the browser: leetdata.ai/problems #SQL #dataengineering

Switching into data engineering? The interview won't stop at one SQL question. Start each one free on leetdata.ai: 1. SQL & PySpark in your browser 2. Data modeling canvas 3. 45-min AI mock interview 4. Kafka & Spark cohorts leetdata.ai #dataengineering

Claude now builds live dashboards from raw data. DEW #292's take: dashboards stay. They're how a company agrees on its numbers. BI tools feel it. Who owns a dashboard nobody wrote SQL for? dataengineeringweekly.com/p/… #bi

5 rows: 'a', '', ' ', NULL, 'b' COUNT(*) = ? COUNT(col) = ? COUNT(NULLIF(TRIM(col), '')) = ? Answer before you run it. leetdata.ai/problems/blank-v… #sql

Parquet is how columnar data sits on disk. Arrow is how it sits in memory. Engines that share Arrow pass data with zero copies, no serialization step. From Hoyt Emerson's overview in DEW #292. Practice: leetdata.ai/paths/lakehouse #dataengineering

eBay monitors 2.6B listings without slowing the fee path. Write a tiny event ID to a local ring buffer. Add 250 columns of context later in ClickHouse. Keep the hot path small. dataengineeringweekly.com/p/… #systemdesign

Data modeling round, one decision per scenario: SCD2: insert, never update Org chart: manager_id points at employees Data vault: hubs, links, satellites Schema evolution: additive only Practise on a canvas: leetdata.ai/modeling #dataengineering