Тестовое заданиеPolza Agency5 Aug 2026
polza - data-cleaning pipeline with an anomaly report
- Next.js 16
- PostgreSQL
- Data cleaning
- SQL
- Vitest
- Docker
A data-cleaning test task. The pipeline pulls 20 pages of company records from an API (1,000 rows that turn out to be 994 distinct companies) plus a 207-row review CSV, normalises and deduplicates them into Postgres - catching 6 CSV rows that are existing companies under new ids, differing only in stripped quotes - and logs every problem it meets into an ingest_issues table. The anomaly report is generated from that table rather than written by hand. A Next.js 16 page serves the cleaned data with search, filters, sorting, pagination, the anomaly log and light and dark themes, with reads cached for five minutes behind a secret-protected revalidate endpoint. Docker Compose for the stack, Vitest for the pipeline.