The search for new energy sources and materials is at the foundations of scientific and technological progress. These two, often interconnected, fundamental elements open a way to create novel designs and improve existing ones for all types of machines and devices. Perovskite solar cells (PSCs) are a great example of this kind of system. Achievements in halide perovskite materials research made a remarkably rapid growth of energy conversion efficiency in PSC technology possible. In turn, a manufacturing process for the technology is potentially cheaper and more straightforward (relative to Si-based). As a result, the PSCs can be a solid background for producing the next generation of solar cells. However, unfortunately, PSCs are still not ready for a market, primarily because of three problems: large-scale production, stability, and toxicity. The first problem is mainly a result of the rawness of the technology, requires mostly resolving engineering tasks, and is not a fundamental barrier. The second problem – stability has a deeper connection to the properties of the material itself. The nature of degradation processes in PSCs is not completely clear. Still, different studies have found some promising ways to stabilize PSCs for a range of thousands of hours. Thus, a combination of engineering and material research can solve this problem, despite its deeper basis. The third problem, toxicity, is a fundamental property of the high efficient PSCs because of its lead-based compound. Lead is one of the critical components of halide perovskites used in the technology, and at the same time, this metal has confirmed toxicity. The water solubility of lead in halide perovskites (primarily as PbI2 and PbBr2) complicates the situation for the whole lifecycle of the devices - from production to recycling. The toxicity of the materials used in PSCs can affect one of the main goals of solar energy: to be produced safe, clean, and sustainable. Taking into account these facts, I focused this thesis on searching for lead-free alternatives for PSCs. Initially, I focused on searching for a model to predict promising vacancy-ordered double perovskites' formability as part of initial screening. I demonstrated the applicability of the geometrical approach based on Goldschmidt's Tolerance Factor for a simple, easy-to-use formability prediction technique for this class of materials. During this part of the study, I also demonstrated a difference between using the method for classic and vacancy-ordered perovskites. In the next part, I used a novel approach based on deep learning to design a new model of perovskite formability. As a result, I successfully forecast perovskite formability with an accuracy of over 98% for the best case. Moreover, I found that the resulting models were able to find some inaccuracies in the original data. Following, I synthesized and researched the structure and properties of tellurium-based vacancy-ordered perovskites. During the research, I found an unexpected complex situation of the optical gap determination for the materials. To solve this challenge, I demonstrated different approaches of absorption spectra fitting and cross-referenced it with other experimental data.