Building physics-aware AI for aerospace predictive maintenance

Michele E. J. Maestrini

MSc Data Science. First-Class BEng Civil Engineering. I bring a structured, systems-based approach to complex, high-stakes problems — and the ability to operate effectively where the path is not already defined.

Technologies evolve. Priorities shift. Organisations are routinely forced to solve problems they have not encountered before. The people who create the most value are those who identify the right problem, test ideas with rigour, and move from uncertainty to clarity at speed. That is how I operate.

FusionCore v0 is the direct expression of that approach: a physics-aware predictive maintenance pipeline for turbofan Remaining Useful Life estimation on NASA C-MAPSS, combining zero-leakage preprocessing, regime-aware normalisation, explainability, and operationally grounded evaluation. The objective was not a model with strong metrics — it was a system that is reliable, interpretable, and deployable in safety-critical environments. XGBoost achieved RMSE 14.85 cycles, NASA Asymmetric Score 4,336.3, and Critical-band F₂ 0.9339 across 707 held-out engines, benchmarked against TFT, N-HiTS, and DeepAR under literature-standard configurations.

My MSc research in few-shot learning, transfer learning, and Bayesian optimisation reinforced a parallel discipline: efficient model design under data constraints. Across both bodies of work, the method is the same — understand the problem deeply, adapt fast, validate rigorously, build systems that hold under real-world conditions.

Aviation does not tolerate unverified claims, untested systems, or complacent engineering — and that is exactly the standard I build to. AI will reshape this industry, but only where it earns its place through explainability, operational trust, and safety-critical rigour. Expertise without restlessness is complacency. I operate as a student of this field — asking the right questions, pushing the boundaries of what teams and systems are capable of, and building work that holds up when the stakes are highest.