Case study
AI Pilot Training Assistant
Backend for an AI-driven pilot training app: speech analysis of spoken answers, assessment tracking and interactive training modules, on a replicated PostgreSQL setup.
- Laravel
- Vue.js
- PostgreSQL
- Laravel Octane
- Pgpool
- Docker
- CI/CD
01 / The problem
The problem
Pilot trainees practise spoken procedures again and again, and instructors need a consistent record of how each trainee is progressing. The app had to take a trainee's spoken answer, analyse it, score it against the module and keep that history available during training sessions, without the database becoming a single point of failure.
02 / My role
My role
Backend Engineer. I designed and built the backend: the API used by the Vue.js frontend, the data model for modules and assessments, the speech-analysis workflow, and the database and deployment setup.
03 / The solution
The solution
A Laravel API running on Laravel Octane, so the app stays responsive while recordings are processed and scored. Interactive modules are built from steps. A trainee's spoken answer is submitted, analysed, scored and stored against the step, so every attempt becomes part of their assessment history. Instructors see progress per trainee and per module. PostgreSQL runs as a replicated cluster behind Pgpool, which spreads read queries across replicas and fails over if the primary goes down. The services run in Docker containers and are released through a CI/CD pipeline.
04 / The result
The result
Trainees get feedback on each spoken answer and their progress is recorded module by module. The replicated database keeps the app available when a node fails, and releases go out through the pipeline instead of by hand.
