Soil microbial communities are particularly sensitive to selenium contamination, which has seriously affected the stability of soil ecological environment and function. In this study, we applied high-throughput 16S rRNA gene sequencing to examine the effects of low and high doses of sodium selenite and the selenite-degrading bacterium, Rhodococcus qingshengii PM1, on soil bacterial community composition, diversity, and assembly processes under controlled laboratory conditions. Our results indicated that sodium selenite and strain PM1 were key predictors of bacterial community structure in selenium-contaminated soils. Exposure to sodium selenite initially led to reductions in microbial diversity and a shift in dominant bacterial groups, particularly an increase in Actinobacteria and a decrease in Acidobacteria. Sodium selenite significantly reduced microbial diversity and simplified co-occurrence networks, whereas inoculation with strain PM1 partially reversed these effects by enhancing community complexity. Ecological modeling, including the normalized stochasticity ratio (NST) and Sloan’s neutral community model (NCM), suggested that stochastic processes predominated in the assembly of bacterial communities under selenium stress. Null model analysis further revealed that heterogeneous selection and drift were primary drivers of community turnover, with PM1 inoculation promoting species dispersal and buffering against the negative impacts of selenium. These findings shed light on microbial community assembly mechanisms under selenium contamination and highlight the potential of strain PM1 for the bioremediation of selenium-affected soils.