3 months ago
Base Salary
$188k - $322k/yr
Responsibilities
- Design, prototype, evaluate, and productionize computer vision and deep learning systems for 3D reconstruction, scene understanding, semantic modeling, and structured property representation.
- Read, reproduce, and adapt academic and industry approaches; train models; design experiments; build proof-of-concepts; and integrate successful methods into production systems.
- Work on mobile capture, aerial imagery, multimodal sensor fusion, multi-view reconstruction, pose estimation, feature matching, dense geometry, model fitting, semantic understanding, and structured 3D outputs.
- Balance research quality, user impact, production constraints, performance, reliability, scalability, and cost when making technical decisions.
- Partner with reconstruction, Product, Design, Engineering, modeling, infrastructure, graphics, frontend, and backend teams to scope experiments and production capabilities.
- At the Staff level, frame problems, size opportunities, prioritize technical investments, define technical direction, and make stop/iterate/scale decisions across reconstruction, data, quality, and product impact.
- Set standards for performance, reliability, maintainability, and cost, and mentor engineers through design reviews, architecture discussions, experiment plans, and production-readiness reviews.
- Communicate technical tradeoffs, resource needs, and roadmap implications to cross-functional partners and leadership.
Requirements
- At least 5 years of experience in computer vision, machine learning, or 3D reconstruction in academic research and/or industry settings.
- Hands-on experience with modern 3D vision or reconstruction, such as multi-view geometry, pose estimation, camera calibration, structure-from-motion, feature matching, dense reconstruction, model fitting, optimization, scene understanding, or semantic modeling.
- Practical deep learning experience applied to 3D reconstruction, geometry, correspondence, pose, segmentation, semantic understanding, VLMs, or related spatial machine learning problems.
- Strong software engineering skills in Python and/or C++.
- Ability to prototype, train models, rigorously evaluate approaches, and translate research into practical systems.
- Experience designing experiments with metrics, baselines, datasets, evaluation plans, and go/no-go criteria.
- Ability to produce project plans, technical specifications, research summaries, experiment reports, and production-readiness documentation.
- Track record of technical ownership and independently driving projects from ambiguity through delivery.
- Strong collaboration, product judgment, and engineering judgment across research, engineering, modeling, infrastructure, product, and other cross-functional teams.
- Master's or PhD in Computer Science, Machine Learning, Computer Vision, Robotics, Graphics, or a related field.
- Staff-level candidates should have experience shipping production CV, ML, or 3D reconstruction systems at scale; defining technical direction and architecture; leading systems from research through deployment and ownership; mentoring engineers; and influencing technical and non-technical stakeholders.
- Preferred qualifications include publications or research contributions in relevant venues, large-scale production ML, model serving, monitoring, MLOps, distributed training, scalable data pipelines, geometry processing, graphics, rendering, CAD/BIM workflows, mesh processing, interactive 3D systems, and AI-assisted development tools.
Benefits
- Comprehensive medical, dental, and vision coverage for employees and dependents.
- Unlimited and flexible vacation policy.
- Generous paid parental and new child bonding leave.
- Monthly mandatory self-care days.
- Fitness classes, meditation and mindfulness tools, virtual therapy, and family planning assistance.
- Support for management training, conferences, workshops, and certifications.
- Full-time position; employees within 50 miles of San Francisco or New York City hubs are expected in the office at least three times per week; fully remote work is not currently available.
Tech Stack
Categories
About Hover
Hover transforms smartphone photos of any property into an accurate, interactive 3D model, giving people the power to unlock the true potential of home. With Hover, homeowners and construction pros can design homes with Generative AI, get better estimates, receive exact measurements, and more. Hover’s insurance industry customers use our integrated claims automation system to reduce costs, decrease cycle times, and streamline the inspections to claims workflow. Offering solutions to homeowners, construction pros, and the insurance market, Hover makes home improvement easier and more efficient than ever before. Every three seconds, a property is captured with Hover. Over 300,000 contractors and home improvement professionals, and over 500,000 homeowners have captured millions of properties and modeled billions of square feet using Hover's technology.