From data engineer to FDE: when the data belongs to the customer, quarantine the bad rows and keep the demo alive
An FDE Times guide for data engineers switching into forward deployed work argues that at a customer site you negotiate with messy ERP and Excel data instead of owning the schema: load what is valid, quarantine the rest with reasons, and turn mappings into reusable product.
FDE Times published a career guide on 13 October aimed at data engineers moving into forward deployed roles. The core shock, it argues, is ownership: on an internal team you define the schema and file tickets upstream; on a customer site the schema already exists, often disagrees with itself, and there is no upstream to blame.
The piece walks through a concrete invoice scenario—SAP export, a finance Excel file, and scanned PDFs—and says the FDE response is not to stop the job when 60 of 1,000 rows fail validation. Load the 940 good rows, quarantine the rest with specific reasons for the business owner to fix, keep retries for infrastructure failures only, and keep mappings in configuration so a second source can reuse the same gate.
That matters for FDEs because postings at companies such as Valdera, Rowspace and Vori already describe messy formats, post-sale integrations, and reconciliation as the day job. The guide also ties the work back to the product loop: harden what customers actually use, and package the best of one engagement so it becomes a feature for the next.
Source: fdetimes.net