Research Robotics/Computer Vision Engineer

About the company Skild AI is building the world's first general-purpose robotic intelligence — robust, adaptive, and designed for unseen scenarios. The team bets heavily on massive-scale, data-driven machine learning to unlock widespread robot deployment across society. They value demonstrated ability and attitude over pedigree. About the role This research engineering role focuses on developing perceptive, intelligent, and adaptable robotic systems with an emphasis on 3D computer vision and autonomous navigation. The engineer will design perception pipelines, optimize SLAM systems, and build learning-based algorithms for real-world robotic control. On-site presence in San Mateo, CA five days per week is required. Responsibilities - Implement perception pipelines for safe robot exploration and navigation in real-world environments - Reconstruct 3D scenes from monocular images, estimate camera poses, and optimize SLAM systems - Develop software tools for vision-only robot localization - Build life-long mapping software via optimally merged pose-graphs - Implement visual servoing relative to detected/tracked objects - Research techniques to handle glare during robotic mapping and navigation - Build data collection pipelines for streaming hand movements to train robot manipulation tasks - Maintain camera and 2D lidar navigation stacks, including bug fixes and new feature deployments Requirements - Master's degree (or foreign equivalent) in Computer Vision, Robotics, or directly related field - 1+ year of experience in Machine Learning or Data Science - Experience with 3D scene reconstruction using monocular video, meshes, point clouds, NeRF, and Gaussian Splats - Experience reconstructing rigid and articulated hand-held objects from video, including hand configuration and relative pose inference - Knowledge of generative computer vision (diffusion models, occlusion handling, data-driven priors) - Experience optimizing attention-based perception models for autonomous navigation - Familiarity with Neural Architecture Search (NAS) for perception backbones - Cloud-based deep learning training experience (AWS, GCP, or Vertex AI) with distributed training and optimized data loading (PyTorch dataloader, sharding, hardware-in-loop) Nice to have - Not specified Benefits & perks - Not specified Compensation - Base salary: $250,000–$300,000 USD per year. Equity and additional benefits not specified in the posting.