Computer Vision / Machine Learning Engineer

Geom is building an AI platform focused on production-scale architecture, helping homebuilders automate repetitive drafting, review, and construction-document workflows. Its technology integrates with the tools and plan sets builders already use, reducing manual effort and improving consistency. Backed by prominent technology investors, Geom's team combines deep expertise in architecture, homebuilding, AI, and startup environments. The company is headquartered in San Francisco and operates at the intersection of built environment and advanced machine learning. Additional information about Geom and its platform can be found at .Scroll down to find an indepth overview of this job, and what is expected of candidates Make an application by clicking on the Apply button.The RoleWe're looking for a Computer Vision / Machine Learning Engineer to help build the intelligence behind Geom.You'll work on challenging problems at the intersection of computer vision, multimodal AI, geometry, machine learning, and architectural data. The goal is not simply to recognize what appears in a drawing, but to understand the elements, relationships, constraints, and intent represented within complex architectural plan sets, and turn that understanding into reliable automation.This is an early-stage role with the opportunity to make an impact. You'll work closely with our engineering, product, and architecture teams to take ideas from experimentation through production and apply them to real-world building plans at scale.What You'll Work OnDevelop models and systems for understanding architectural drawings, floor plans, construction documents, and other complex visual information.Build computer vision and multimodal pipelines for detecting, classifying, extracting, and reasoning about architectural elements and their relationships.Combine vision models, foundation models, geometric reasoning, and structured representations to create semantic understanding of plan sets.Develop methods for matching and tracking elements across drawings, sheets, revisions, and plan variations.Work on problems involving object detection, segmentation, document understanding, spatial reasoning, image registration, vector geometry, and visual similarity.Design training, evaluation, and data-generation pipelines that allow models to improve rapidly as we encounter new plan types and workflows.Develop practical approaches for working with imperfect, noisy, inconsistent, and highly variable real-world architectural data.Translate experimental ML capabilities into reliable production systems.Establish evaluation frameworks and metrics for model accuracy, reliability, and downstream automation quality.Collaborate directly with domain experts to understand architectural conventions and convert that expertise into scalable technical systems.What We're Looking ForStrong experience building and deploying machine learning or computer vision systems.Deep understanding of modern computer vision techniques and neural network architectures.Experience with one or more areas such as object detection, segmentation, document AI, multimodal models, vision-language models, image matching, geometry processing, or spatial reasoning.Strong Python skills and experience with modern ML frameworks such as PyTorch.Experience building datasets, training pipelines, evaluation systems, and production inference workflows.Strong software engineering fundamentals and the ability to turn research ideas into maintainable production systems.Comfort working on ambiguous problems where the right technical approach may need to be invented.Ability to move quickly between experimentation, implementation, debugging, and deployment.Excitement about working closely with a small, highly collaborative team in an early-stage environment.You may be particularly well suited for this role if you have experience with:Document understanding or document intelligenceCAD, architectural drawings, engineering drawings, maps, diagrams, or other structured visual documentsVector graphics, computational geometry, graph representations, or spatial dataOCR and layout understandingImage registration, feature matching, or correspondenceNumerical optimizationSynthetic data generation and data augmentationFine-tuning or adapting foundation modelsModel evaluation and benchmarking for complex reasoning tasksML infrastructure and large-scale inference systemsWhy GeomYou'll have the opportunity to work on AI problems that are both technically difficult and immediately connected to the physical world. Architectural plans contain an enormous amount of structured information, but much of the work required to interpret, modify, and produce them remains manual. We're building the intelligence layer that changes that. At Geom, you'll have meaningful ownership from the beginning, work directly with experienced founders and domain experts, and help define a new technical category at the intersection of AI, architecture, and homebuilding.Don't See a Perfect Match?Great people rarely fit perfectly into a checklist. If you're excited about computer vision, multimodal AI, geometry, or building intelligent systems that understand the built environment, we'd like to hear from you.CompensationPursuant to applicable state and local pay transparency laws, the anticipated salary range for this position is $160,000–$220,000 annually. Actual compensation will be determined based on factors including relevant experience, skills, education, certifications, geographic location, and internal equity. xhyhwjd BenefitsHealth and dental insurancePaid time off and holidaysCollaborative and innovative work environmentRemote work options (where applicable)Geom is an equal opportunity employer. We welcome applicants of all backgrounds and make employment decisions based on skills, experience, and what each person can bring to the team, without regard to race, color, religion, sex, sexual orientation, gender identity or expression, national origin, age, disability, veteran status, or any other characteristic protected by law.