Line maintenance planning must absorb constant change. A production plan depends on work packages, flight operations, ground transit times, and which technicians are available and qualified at any given moment, and all those inputs change constantly. When a single flight is delayed or cancelled, the plan must often be rebuilt, rapidly.
Planning well under those conditions, with dynamic, incomplete information, is a genuinely hard computational problem. The difficulty lies not in understanding individual decisions, but in orchestrating thousands of interdependent decisions: which technician performs which task and when, given their qualifications, availability, and precedence constraints of individual tasks. Human planners rely on heuristics and intuition. Computers harness mathematics.
Saurabh Fadnis, our colleague at QOCO, is an expert in this field of constraint programming and optimisation. Saurabh earned his PhD in Computer Science from Aalto University on reductive approaches to automated planning with partial observability: how to take messy, uncertain planning problems and transform them into forms that can be digested by constraint programming.
Building powerful software solutions for aviation requires understanding and modeling complex sets of underlying drivers. The difference between surface-level tools that produce plans that look plausible and a solution that is built on rock-solid mathematical foundations that mirror operational reality is rigour. The fact that plans hold up during sudden changes and rare edge cases means planners can trust the software.
This is the design philosophy behind QOCO Assignment, our agentic AI and ML-powered solution for maintenance optimisation. It also reflects our talent strategy: we hire the very best people with experience in solving hard problems, because our customers deserve more than surface-level answers.
Link to research: Aalto - Public Defence Msc Saurabh Fandis