Engineering programmes generate large volumes of interconnected requirements, verification records and design changes across different phases of avionics development.
My contribution
I independently designed and built the original Python server-client prototype, integrating the engineering data sources and automating consistency checks, traceability tracking and process monitoring.
Turn extensive engineering activities into a coherent picture
The starting point was a heterogeneous collection of engineering records, continuously updated by multiple specialists in DOORS and PVCS. These tools were not designed to provide an up-to-date overall picture, lacking automated mechanisms to identify and notify users of inconsistencies, gaps and unresolved issues.
From prototype to enterprise tool
The prototype secured internal support for industrialisation and became an official Leonardo tool, subsequently adopted across multiple helicopter programmes. Engineers, testers and managers now receive notifications tailored to their responsibilities, accelerating reviews and enabling earlier detection of inconsistencies.
Add AI assistance grounded in data
I am now exploring an on-premise AI assistant with an integrated knowledge retrieval layer. A local prototype combines Qwen, Ollama and Graphify to provide the LLM with relevant engineering procedures and technical knowledge. The goal is to extend WIZARD with context-aware AI assistance grounded in reliable data sources.