Operational spare parts planning in maintenance, repair, and overhaul requires interventions under uncertainty about repair completions, demand, and contractual obligations. We study this as a finite horizon Markov decision process in which a planner with real time pipeline visibility can expedite repair jobs, lease inventory under short and long term contracts, and buy emergency stock over a short horizon. Prior operational models consider these interventions in isolation. We jointly optimize all three and, to our knowledge, are the first to incorporate leasing. We solve the model exactly using backward recursion, and approximately via proximal policy optimization where exact methods are infeasible. We validate the approach on an aerospace spare parts case study. Three findings emerge. First, job specific expediting conditioned on repair progress and contract type yields a richer policy than aggregate models: under part repair contracts it follows a just-in-time structure set by SLA due dates and acts as a lever on demand, not supply alone. Second, the interventions are complementary: expediting dominates, short term leasing absorbs mild shortages without the commitment of buying, and buying is justified only under sustained demand. Expediting and leasing together cut expected cost by 48% to 64%. Third, the approximate policy stays close to optimal across tractable instances, with errors traceable to operational regime. For practice, the strong state dependence of the optimal policy means static intervention rules forgo substantial savings, while leasing, though largely overlooked in the literature, emerges as a high value lever that managers should weigh alongside expediting and buying.