2026-05-18 · 5 min read · Rahul Chitturi
- data
- engineering
Spreadsheets are the universal DSL of business teams. Engineering receives CSV exports with merged cells, stray formulas, and locale-specific decimals. Normalizing early prevents silent corruption downstream.
Clean before convert
JSON arrays of objects are easier to validate in CI than raw CSV.
- Remove summary rows and blank lines
- Standardize column headers to snake_case
- Fix European decimal commas
- Validate numeric columns
Into MATLAB or Simulink
Numeric matrices from JSON feed lookup tables and test vectors. Keep the JSON in git as the canonical calibration source; generate MATLAB literals locally when models need them.