Computer Vision Engineer

Computer Vision EngineerEfference - Robust Robotic PerceptionLocation: San Francisco, CAEmployment Type: Full TimeLocation Type: On-siteDepartment: EngineeringWe are seeking an experienced Computer Vision Engineer focused on spatial AI, algorithm optimization for edge hardware, and building robust 3D perception systems. This role blends VIO/VSLAM development, model optimization, and edge compute deployment to enable real-time robotic vision. Your work will center on ensuring complex spatial models run efficiently on dedicated NPUs, delivering high-performance, low-latency perception directly on the device.RequirementsExperience with 3D computer vision, Visual Inertial Odometry (VIO), or VSLAM systems.System-level understanding of perception pipelines, from raw sensor input to real-time spatial mapping.A strong focus on delivery, with a track record of deploying robust algorithms to edge environments.Strong C/C++ and Python programming skills.Familiarity with deploying and accelerating machine learning models on edge devices and dedicated Neural Processing Units (NPUs).ResponsibilitiesDevelop and deploy robust VIO/VSLAM and 3D perception algorithms to run directly on-device.Optimize vision pipelines to maximize inference speed and efficiently leverage embedded compute resources.Collaborate with infrastructure and hardware teams to minimize end-to-end latency across the entire perception stack.Test, benchmark, and validate perception systems on physical lab robots in real-world environments.Diagnose and aggressively resolve tracking drops, algorithmic failure modes, and edge cases to ensure highly reliable spatial awareness.Nice to HaveExperience with model distillation, quantization, and edge AI optimization techniques.Familiarity with modern Vision Foundation Models and their deployment.Background working with stereo depth estimation or multi-sensor fusion (e.g., tightly coupled Camera + IMU systems).