This page displays an extensive list of my research work
FlowCast: Advancing Precipitation Nowcasting with Conditional Flow Matching
Summary: Developed the first end-to-end probabilistic model leveraging Conditional Flow Matching (CFM) for precipitation nowcasting. Achieved state-of-the-art accuracy on SEVIR/ARSO benchmarks while improving inference efficiency by 100x against Diffusion (DDIM) through optimized sampling (1–10 steps).
EdgeEar: Efficient Ear Recognition for Edge Devices
Summary: Modified the EdgeNeXt CNN-Transformer architecture using Low-Rank Linear (LoRaLin) layers to achieve a 50x parameter reduction. Obtained the lowest Equal Error Rate (EER) on the UERC2023 dataset.