The fast development of quantum computing is a major threat to the classical cryptographic systems and the emergence of secure and efficient post-quantum cryptographic (PQC) is required. Lattice-based cryptography is one of the potential solutions that have been adopted as one of the promising solutions due to its excellent security grounds and quantum resistant nature. Nevertheless, there are difficulties of large key sizes, high computing power and flexibility to diverse application environments that plague existing lattice-based schemes. This paper introduces a new model which combines lattice-theoretic concepts and discrete optimization methods to improve the performance and scalability of PQC systems. Its methodology includes the use of optimization-based key generation, adaptive parameter choice, and using efficient encryption-decryption mechanisms that are based on the Learning With Errors (LWE) problem. Key metrics, such as key size, computation time and error rate, are used to evaluate the comprehensive performance and compared to the known schemes, such as Kyber, NTRU and Saber. Experimental evidence shows that the given model has the smaller key size of 2.4 KB, shorter calculation time of 13.9 ms and smaller error rate of 0.3 that is superior to existing techniques. The results suggest that the suggested framework offers the best compromise in terms of security, performance, and scalability, thus it is applicable to the next-generation secure communication framework in the post-quantum world.