We present a Semiempirical Pseudopotential Method for accurately computing the band structures and Bloch states of monolayer transition-metal dichalcogenides (TMDCs), including MoS2, MoSe2, WS2, and WSe2. Our approach combines local and non-local pseudopotentials, carefully fitted to replicate fully self-consistent density-functional theory results, while using only a minimal set of empirical parameters. By expressing the total potential as a sum of a few separable components, we achieve both accuracy and computational efficiency. The resulting pseudopotentials are transferable to more complex systems such as bilayer, trilayer, and moire superlattices, offering a reliable foundation for large-scale simulations. When integrated with artificial intelligence techniques, this method provides a powerful tool for the design and exploration of TMDC-based nanodevices for next-generation optoelectronic applications.