In surveillance of cellular energy homeostasis, lipid droplets (LDs) dynamically interact with other organelles, such as mitochondria, to tightly engage in metabolic regulations. The transient nature of LDs’ interactions, however, poses technical challenges to snapshot the enriched information by classical fractionation methods. Herein, we develop a small molecule based photocatalytic protein proximity labeling method (namely LipoID) to in situ label, capture and profile LDs’ interacting proteome in live cells. To this end, a set of LD-targeting probes are designed to undergo photocatalytic protein modifications in close proximity to LDs by nucleophilic substrates. Photochemical mechanistic studies confirm predominant modification on histidine residues via type I radical pathway. We show the LipoID method identifies inter-organellar interactions, particularly with mitochondria, besides reliable capture of validated LDs’ biomarkers (e.g. Plins). We also exemplify LipoID coupled with comparative proteomic analysis discovers key mitochondrial channel VDAC3 to regulate the inter-organellar distance between LDs and mitochondria. Together, the small molecule based LipoID method reported herein allows us to in situ and de novo study LD interactions without genetic-encoding a functional tag to a known biomarker.