Build and ship production-scale computer vision models end-to-end: data collection, dataset curation, training, deployment (cloud and edge), monitoring, and iteration. Work with real roadside hardware and sensors to deliver efficient, accurate detection, segmentation, classification, embeddings, and re-identification models at scale.

Build and ship computer-vision ML models end-to-end: data collection and curation, training, deployment to cloud and edge, monitoring, and iteration. Work with real roadside hardware and large-scale vision pipelines (detection, segmentation, classification, embeddings, re-identification). Choose between classical CV and SOTA models and optimize for accuracy, speed, and power.