Decreasing defects, waste time, meeting customer demand and being adaptable are the goals of a Zero Defect Manufacturing (ZDM) strategy. Scheduling is an important tool to perform that. It should take in account buffer size allocation. In this study, a method to solve the Buffer Sizing Problem (BSP), which is NP-hard problem. The current research work focuses on finding the optimal buffer allocation using Tabu-search (TS) algorithm. The goal is to minimize buffers' sizes while maintaining a certain productivity. The evaluation of the alternative buffer solutions were performed using the following performance indicators; Makespan, Tardiness and the Buffers Cost. In the developed method the following are considered: multitasking machines subjected to non-deterministic failure, non-homogeneous buffer sizing, and non-sequential production line. The propose approach was tested via a real life industrial use case from a leading Swiss company in high precision sensors. The simulation results showed that the proposed methodology can effectively design the buffer strategy for complex production lines.