In the last decade, blockchain technology has undergone a strong evolution. The maturity reached and the consolidation obtained have aroused the interest of companies and businesses, transforming it into a possible response to various industrial needs. However, the lack of standards and tools for the development and maintenance of blockchain software leaves open challenges and various possibilities for improvements. The goal of this thesis is to tackle some of the challenges proposed by blockchain technology, to design and implement analysis, processes, and architectures that may be applied in the real world. In particular, two topics are addressed: the verification of the blockchain software and the code optimization of smart contracts. As regards the verification, the thesis focuses on the original developments of tools and analyses able to detect statically, i.e. without code execution, issues related to non-determinism, untrusted cross-contracts invocation, and numerical overflow/underflow. Moreover, an approach based on on-chain verification is investigated, to proactively involve the blockchain in verifying the code before and after its deployment. For the optimization side, the thesis describes an optimization process for the code translation from Solidity language to Takamaka, also proposing an efficient algorithm to compute snapshots for fungible and non-fungible tokens. The results of this thesis are an important first step towards improving blockchain software development, empirically demonstrating the applicability of the proposed approaches and their involvement also in the industrial field.