Top Computer Vision Jobs in San Francisco Bay Area, CA

Create precise vector annotations and segmentation masks on geospatial imagery, validate and correct automated detections, maintain documentation and QA, support dataset pipelines, and assist ML teams with inference and model evaluation to produce production-ready training data.

As a Research Engineer, you'll enhance visual understanding capabilities by training and deploying models to make video datasets queryable, driving down costs, and collaborating on data taxonomy design.

The Computer Vision AI & ML Engineer will develop and optimize deep learning models for robotics perception systems, collaborating across teams to integrate models and evaluate performance.

Develop robot perception systems for human pose estimation, 3D vision, sensor calibration, visual SLAM, multimodal synchronization, and 6DoF pose estimation. Train and optimize perception models for embedded deployment using TensorRT and CUDA, build automated annotation and quality-assurance pipelines, and connect perception accuracy with robot training outcomes. Collaborate with robotics, machine learning, and data infrastructure teams to deliver high-quality annotations for teleoperation and embodied AI systems.

Perception Interns will experiment with and evaluate machine learning solutions for onboard and offboard drone perception. They will train models using large real-world, simulated, and internet-scale datasets; write production code for model training, validation, and integration; research novel perception methods; optimize models for deployment; and collaborate with planning, navigation, mapping, and hardware teams.

Conduct research and engineering to improve Claude's visual and spatial capabilities across pretraining, SL, and RL. Run experiments on architectures and data strategies, build agentic tools and evaluation benchmarks, develop data ingestion pipelines, and partner with product to deploy vision improvements for real-world API use cases.

As an ML Engineer at Mach9, you will design and train computer vision and 3D ML models for extracting features from LiDAR and imagery, owning models through the full product lifecycle and collaborating with teams for deployment.

Build and deploy real-time multimodal perception systems (camera, LiDAR, radar) for 3D/BEV scene understanding. Optimize models for low-latency embedded inference, design auto-labeling and data pipelines, evaluate metrics, and validate systems on real machines in rugged environments.

Own and develop the end-to-end robot perception stack and high-reliability manipulation policies. Research, deploy, and iterate transformer/diffusion and other learned policies on real hotel robots, handle multimodal sensor data, debug field failures, and optimize models for real-time (~10 Hz) inference on constrained edge hardware.

Develop and productionize photogrammetry and radar image-formation pipelines for large-scale 3D reconstruction. Improve pipeline stages (bundle adjustment, dense matching, point clouds, geo-referencing), optimize storage and databases for 3D geospatial data, and contribute to data infrastructure, analytics, and delivery systems for satellite-derived products.