Computer Vision Research Engineer

Responsibilities

• Build SOTA CV models for detection, segmentation, OCR, and reasoning over construction drawings
• Own end-to-end pipelines including data ingestion, labeling, training, evaluation, and low-latency inference
• Scale model performance regarding accuracy, robustness, and latency across diverse drawing formats
• Integrate models into production features working closely with fullstack and product teams
• Analyze customer data to identify and fix model failure modes
• Apply pragmatic research to implement state-of-the-art techniques quickly

Requirements

• Deep expertise in Computer Vision and Deep Learning (segmentation, detection, document AI, vision-language)
• Proven experience putting ML models into real production environments with SLAs
• Ability to work with messy, real-world data (scanned, handwritten, or distorted drawings)
• Strong execution focus with the ability to ship code based on research
• Ability to work full-time in-person in the SF Bay Area

About the Company

Bobyard automates construction takeoffs by solving non-trivial computer vision problems, replacing manual processes with high-accuracy models that work on messy, real-world drawings.