Machine Learning Engineer, Predictive Maintenance

About the role: Trakt System, Kquika's flagship predictive maintenance platform, sets the standard for AI in aviation maintenance: 95% prediction accuracy, 2 to 13 weeks of advance warning before component failures, and measurable results for operators including roughly 22% lower maintenance costs and 40% fewer aircraft-on-ground events. Behind those numbers is a suite of six best-in-class AI models working together, from the Early Warning System that catches unusual patterns before they become problems, to Failure Prevention, Component Life Tracking, Smart Scheduling, Fleet Optimization, and Parts Forecasting. As a Machine Learning Engineer on this team, you will advance the intelligence at the heart of the platform. Your work will determine when an airline grounds an aircraft, orders a part, or schedules a maintenance visit, so accuracy, calibration, and trustworthy uncertainty are the craft here, not afterthoughts. Airlines, government operators, and MROs across Boeing, Airbus, Bombardier, and Embraer fleets rely on these predictions every day. If you want your models making high-stakes operational decisions in the real world, this is that role. Due to US government contract and export control requirements, applicants must be US citizens.

What you will do: Advance the accuracy and reliability of Trakt's six AI model suites, keeping them best in class across the industry Improve component life estimation so operators replace parts at the right time, not too early and not too late Strengthen the Early Warning System's ability to detect subtle anomalies across flight telemetry, sensor data, and maintenance records Refine failure prediction confidence and advance-warning windows that maintenance planners act on directly Expand model coverage across aircraft systems, from air conditioning and electrical power to landing gear and power plant Enhance the explainability layer so maintenance engineers can understand and trust every prediction Contribute to the continuous learning loop where real-world scheduling outcomes improve future predictions Evaluate models rigorously against historical fleet data and monitor performance in production

What we are looking for: US citizenship (required for government contract and export control compliance) 4+ years building and shipping machine learning models to production Deep experience with time-series modeling on sensor or telemetry data Background in reliability modeling, remaining-life estimation, or predictive maintenance Ability to quantify and communicate prediction uncertainty clearly to non-specialists Comfort validating models against sparse, imbalanced real-world failure data Experience monitoring, retraining, and maintaining models in production environments Aviation, aerospace, industrial, or other safety-critical domain experience is a strong plus