The 2025 Summer Intern position is located in South San Francisco, on site, within the Regev Lab at Genentech. This internship focuses on cellular deconvolution, a technique used to infer the unknown composition of bulk RNAseq samples. Traditionally, this method has been limited to estimating abundances of discrete cell types. However, recent approaches utilizing maximum entropy models and generative models show promise in inferring continuous processes such as differentiation. The Regev lab aims to develop a deep generative model that builds on these approaches, trained on extensive and heterogeneous reference atlases containing tens of millions of single cells from hundreds of studies. This model will facilitate the routine profiling of large volumes of bulk RNAseq data collected in-house, with an assessment of its accuracy in epithelial, cancer, and other tissue samples containing diverse and continuous cell states.