Top Computer Vision Jobs in San Francisco Bay Area, CA

Design, train, fine-tune, evaluate, and deploy vision, vision-language, and vision-language-action models for autonomous systems. Build scalable data pipelines, SFT workflows, evaluation frameworks, and optimize models for embedded hardware (ONNX, TensorRT). Collaborate with researchers and engineers to translate ML research into robust, production-ready perception capabilities for defense platforms.

The Senior Staff Machine Learning Scientist will lead research initiatives in computer vision and multimodal understanding, develop novel models, and translate research into practical applications while mentoring team members.

Develop, train, and optimize computer vision and ML models (particularly video-based safety) using petabyte-scale camera and sensor data. Deploy and optimize models for backend and edge inference, collaborate with cross-functional teams, and translate customer needs into scalable ML/CV solutions.

Lead the design, deployment, and optimization of production photogrammetry and 3D reconstruction systems. Drive computer vision architecture, integrate emerging reconstruction and machine learning approaches, improve system performance and reliability, and oversee technical investigations. Establish engineering and observability standards, mentor engineers, guide technical direction, and expand AI coding workflows. The role requires deep expertise in 3D perception, production engineering, C++, machine learning frameworks, and cloud infrastructure.

Leads Zoox’s perception data organization and data flywheel, connecting fleet signals, data discovery, curation, annotation, model training, evaluation, and performance measurement. Manages a multidisciplinary team of data science, data engineering, and labeling professionals. Drives automated annotation, intelligent mining of rare and safety-critical scenarios, log selection and storage, and rigorous tradeoffs among data quality, cost, speed, and scale. Partners with machine learning, metrics, evaluation, and infrastructure teams to improve autonomous-driving models.

Computer Vision Intern developing detection, segmentation, and classification models for food and material recognition. Responsibilities include curating and QAing image/video datasets, building reproducible training and experiment-tracking pipelines, evaluating models, documenting findings, and collaborating with engineering and data teams to deploy computer vision systems to production devices.

Develop and deploy applied ML and computer vision models to understand technical drawings and specifications. Work on multimodal reasoning across diagrams, text, and structured data, translate research into production within ~1 year, present prior research, and collaborate closely with founders to build customer-impacting solutions.

Productionize computer vision and 3D reconstruction systems for embodied AI. Own spatial output quality, robustness, evaluation infrastructure, performance, throughput, cost, and reliability. Diagnose real-world failures involving calibration, scale, coordinate frames, and scene complexity. Collaborate with research, product, engineering, and customers to deliver production-ready capabilities, workflows, and interfaces for simulation, training, evaluation, and deployment.

Hands-on machine vision applications engineer responsible for lab testing, camera and optical setup, proof-of-concept builds, technical reporting, technology evaluation, customer-site troubleshooting, and deployment support. The role also includes image collection, labeling, model training, model evaluation, camera calibration, hardware evaluation, and building internal test resources. Candidates need machine vision fundamentals, AI vision familiarity, troubleshooting ability, and comfort working with hardware, industrial automation, and manufacturing customers.

Develop multimodal AI systems for image, video, generative AI, and virtual try-on applications. Optimize diffusion models for controllability and efficiency, design visual evaluation frameworks, productionize research prototypes, and collaborate with infrastructure teams to improve model training and inference.