Pharmacological cancer therapy can lead to a range of outcomes, even among patients with cancer in the same location, originating from same cell type, and with the same biomolecular markers. Treatment heterogeneity is also reflected at the cellular level, where cells within an individual tumor can exhibit discordant responses to a drug. This fractal nature of cancer diversity has encouraged the development of hundreds of unique drugs and therapeutic regimens, with the ambition of matching each patient to an ideal treatment option. The rich space of possible treatments is infeasible to study en masse through clinical trials, motivating the use of disease models to screen and hone in on promising candidates. Drug screening must not only explore the large treatment space, but also extract rich measurements for each condition to reflect nuanced pathophysiological processes. These two aspects often pull experimental designs in opposite directions, constraining assays to explore few conditions in detail (high-content analysis) or many conditions superficially (high-throughput screening). Here, I describe the development, validation, and exploration of three technologies for high-throughput and high-content drug screening. The first is a deep learning-powered software tool, OrganoID, that analyzes microscopy images of patient-derived cancer organoids to follow changes in organoid number, size, and shape in response to drug exposure. The second technology, PicoScreen, enables computer-controlled screening of reagent combinations in picoliter droplets with microscopy. Finally, in Screen-seq, a novel molecular barcoding and microfluidic strategy measures transcriptomic responses of thousands of individual cells to all possible combinations of multiple reagents under study. These technologies aim to make screening methods more efficient, comprehensive, and information-rich, to expand our understanding of disease processes, such as cancer, and improve treatment options and selection for patients.