Data Engineering at KPMG

The trade KPMG offers a data engineer is exposure breadth in exchange for depth. You rotate through client problems fast enough to see a wide range of industries and data maturity levels, but you rarely own a pipeline long enough to watch it age into something you'd call production-hardened. Pay lands slightly below relative to the other companies we track, which is consistent with consulting's general positioning against product company comp. The Glassdoor rating sits at 3.7, a little below the middle of the pack, and sentiment on Blind is mixed. Happiness reads as stressed and trending down, which tracks with consulting's normal tension: interesting work, demanding hours, satisfaction that depends heavily on which account you land on. The firm's brand carries real weight for engineers who want to move into staff or principal roles at clients later.

KPMG's data engineering work is shaped by its client-services model: you're not building a single internal pipeline, you're building pipelines for banks, insurers, governments, and manufacturers, each with its own compliance surface, latency tolerance, and legacy stack. That means the data problems are rarely uniform. One engagement might be a financial-sector migration onto Databricks with strict data residency rules; the next might be a batch ETL modernization for a public-sector client still running on-prem. The visible stack, Azure, Databricks and Spark with SQL, Python and PySpark, reflects what enterprise clients actually run in 2026, not greenfield choices. Day-to-day, you'll context-switch more than engineers at product companies, and the work often means inheriting someone else's schema decisions rather than setting your own.