About the job
Perception-focused MLE role dedicated to multi-modal sensor fusion, multi-task deep learning architectures for vehicle semantics and signal detection, dynamic high-resolution vision backbones, and large-scale automated data engines.
Responsibilities
Architect, train, and optimize multi-task deep learning models (PyTorch / JAX) across multi-modal sensor streams.
Build automated data mining pipelines, active learning loops, hard-example curation, and auto-labeling systems.
Develop high-resolution vision architectures and spatial-temporal transformer backbones.
Leverage multimodal foundation models for automated data curation, synthetic edge-case generation, and failure triage.
Profile and optimize models for efficient onboard accelerator inference.
Qualifications
Minimum
2–5+ years training and deploying production vision or multi-modal deep learning models.
Experience with multi-modal sensor fusion (Camera + LiDAR + Audio), multi-task learning (MTL), transformer architectures, and PyTorch / JAX.
Experience building large-scale data curation pipelines, active learning loops, and auto-labeling systems.
Fluency with modern AI developer tools and foundation model workflows for fast prototyping.
Preferred
No preferred qualifications listed.