Priya's Thursday, two ways

Priya, a data engineer on a retail analytics team.

Priya starts Thursday with a failed pipeline. The nightly load from orders_raw stopped after 2:15 AM, and the error log mentions a type mismatch on order_total. She opens ChatGPT and uses the Pipeline Monitoring prompt: she pastes the log lines, explains the source table, and asks for likely causes and a safe fix. It points to a decimal column that arrived as text.

Before lunch, she needs to clean a new returns table. She uses the Data Transformation prompt in Claude, giving column names like return_id, order_date, and reason_code. Claude drafts a small SQL block to trim spaces, cast dates, and drop duplicate rows. Priya tests it on a sample and adjusts one rule.

Then she runs the Data Quality Checks prompt in Gemini. She asks for checks on missing order_id, negative return amounts, and dates in the future. Gemini suggests a query that counts each issue by day. Priya adds two checks to the pipeline and writes a short note for the analysts.

With the pipeline fixed and checks in place, Priya uses the Query Performance Tuning prompt in ChatGPT for a slow analyst query. She pastes the query plan and asks why it scans the full sales table. The suggestion to add a date filter and a composite index helps her cut the wait. She finishes early and uses the time to walk through the data dictionary with a new analyst.