Practice hub
Search SQL, Python, PySpark, production scenarios, and system design practice. Start with a guided section, then narrow the library only when you need it.
Practice items
496
Free starters
116
Modes
5
Recommended for you
Open a browser lab, production scenario, or architecture exercise.
Practice SQL interview problems with seeded tables and browser-side SQL execution. No backend judge required.
Practice Python coding questions in a browser Pyodide runtime with sample tests and model explanations.
Fix production PySpark code without a Spark cluster: reason through DataFrame logic, joins, skew, files, and safe writes.
Debug production incidents with sample data, logs, broken logic, and feedback.
Practice data platform design decisions and defend your trade-offs.
Guided selection
Build a customer-level revenue result with exactly one row per customer. Include only completed orders, return customer_id, customer_name, and completed_revenue, and make sure duplicate status rows cannot inflate the dashboard.
Return every active customer. For customers who clicked campaign SPRING_26, show their latest click timestamp. For customers with no click, keep the customer row and return NULL for last_click_at.
Return one row per order with paid_amount, refunded_amount, and net_revenue. Aggregate each child table to order_id before joining so payment and refund rows cannot multiply each other.
Guided selection
Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.
Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.
Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.
Guided selection
Return one row per order with paid_amount, refunded_amount, and net_revenue. Aggregate each child table to order_id before joining so payment and refund rows cannot multiply each other.
Diagnose the performance issue from logs.
Find why green status is misleading.
Library results
The first 24 are shown to keep this page useful instead of overwhelming.
Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.
Return exactly the requested result set. The browser will compare your output with the model query. After you submit, the browser runs your SQL against the visible sample plus an additional edge-case dataset.
Return exactly the requested result set. The browser will compare your output with the model query. After you submit, the browser runs your SQL against the visible sample plus an additional edge-case dataset.
Return exactly the requested result set. The browser will compare your output with the model query. After you submit, the browser runs your SQL against the visible sample plus an additional edge-case dataset.
Return exactly the requested result set. The browser will compare your output with the model query. After you submit, the browser runs your SQL against the visible sample plus an additional edge-case dataset.
Return exactly the requested result set. The browser will compare your output with the model query. After you submit, the browser runs your SQL against the visible sample plus an additional edge-case dataset.
Write a LEFT JOIN query returning customer_id, customer_name, and city. Your query will be tested with customers that have no profile and with orphan profile rows.
Write a LEFT JOIN query returning customer_id, customer_name, and city. Your query will be tested with customers that have no profile and with orphan profile rows.
Write a LEFT JOIN query returning customer_id, customer_name, and city. Your query will be tested with customers that have no profile and with orphan profile rows.
Write a query that returns customer_id and customer_name for customers with no completed orders. Your query will be tested with NULL customer IDs in the fact table and customers with only cancelled orders.
Write a query that returns customer_id and customer_name for customers with no completed orders. Your query will be tested with NULL customer IDs in the fact table and customers with only cancelled orders.
Write a query that returns customer_id and customer_name for customers with no completed orders. Your query will be tested with NULL customer IDs in the fact table and customers with only cancelled orders.
Write a query that returns customer_id and customer_name for customers with no completed orders. Your query will be tested with NULL customer IDs in the fact table and customers with only cancelled orders.
Return every active customer. For customers who clicked campaign SPRING_26, show their latest click timestamp. For customers with no click, keep the customer row and return NULL for last_click_at.
Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.
Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.
Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.
Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.
Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.
Return country_name, population, and area_sq_km where population is at least 25,000,000 or area is at least 1,000,000. Your query will be tested with rows that qualify by only one threshold.
Return country_name, population, and area_sq_km where population is at least 25,000,000 or area is at least 1,000,000. Your query will be tested with rows that qualify by only one threshold.
Return country_name, population, and area_sq_km where population is at least 25,000,000 or area is at least 1,000,000. Your query will be tested with rows that qualify by only one threshold.
Return country_name, population, and area_sq_km where population is at least 25,000,000 or area is at least 1,000,000. Your query will be tested with rows that qualify by only one threshold.
Return country_name, population, and area_sq_km where population is at least 25,000,000 or area is at least 1,000,000. Your query will be tested with rows that qualify by only one threshold.