Extracting business rules from a legacy system

The context

At Engie GEMS, a legacy application used for market operations had become difficult to maintain. Its replacement involved several specialised applications.

Sole maintainer for the UK and short-term trading scopes of a legacy Python application exceeding 600,000 lines.

The work

Designed an application to manage booking rules for several markets. Cleaned and migrated the rules from the existing system, managed write permissions and implemented cache invalidation for three consuming applications to trigger recalculation when rules changed.

For another application, implemented a mechanism allowing authorised users to edit business rules in an Excel file, which are then validated, converted to YAML and loaded into a Python engine. Changes are logged.

Further work

Built a complete proof of concept for integrating positions with an external provider: transformations, mappings, checks, tests and a timestamped history of exchanges. The proof of concept was validated and selected for industrialisation in October 2026.

Environment: Python, FastAPI, PostgreSQL, MongoDB, Docker, Azure Pipelines.