Automated Project Workflow
An internal platform at OneSubsea that verifies documentation, generates about 95% of project code, and tests it, cutting iterations with customers.
- ~95%of project code generated
- Days → hoursfor interface testing
- 3.5 yrsin production and growing
The problem
Every subsea control system project at OneSubsea started from a stack of documentation, and most of the PLC code and customer documents were derived from it by hand. That meant errors crept in at every step and were only found late, in long review loops with the customer.
What I built
Over three and a half years I built a set of internal tools that grew into one platform covering the whole project workflow. The backend is Python, the UI is Flutter, and everything reports back in Excel so nobody had to learn a new program.
Document verification
A tool that validates multiple project documents and cross-references details between them, notifying owners of what needs to change and proposing edits. It works across all existing projects, so every new check benefits every project at once.
Code generation
Once the documentation is verified, the tool generates around 95% of the PLC project code plus the supplementary documentation delivered to customers.
Automated testing
The generated PLC code is tested directly against the documentation and the results are presented visually. A separate interface tester runs automated commands against the hardware from an external computer, reducing testing time from days to hours.
Document generation
Existing Word deliverables were turned into standardised templates by replacing project-specific data with variables. Documents are now generated on demand as soon as the project data is known, saving multiple departments many hours.
Result
Earlier error detection, far fewer iterations with customers, and faster, more reliable deliveries. The approach freed up engineers for the work that actually needs them.
