Senior Machine Learning R&D Engineer – Matterport

Job Description

CoStar Group (NASDAQ: CSGP) is a leading global provider of commercial and residential real estate information analytics and online marketplaces. Includedin the S&P 500 Index CoStar Group is on a missiontodigitizetheworldsrealestateempoweringallpeopletodiscoverpropertiesinsightsand connections that improve their businesses and lives.

We have been living and breathing the world of real estate information and online marketplaces for over 35 years giving us the perspective to create truly unique and valuable offeringstoour customers. Weve continually refined transformed and perfected our approach to our business creating a language that has become standard in our industry for our customers and even our competitors. We continue that effort today and are always working to improve and drive innovation. This is how we resources available we provide an invaluable edge in real estate.

AboutMatterport:

Matterport is leading the digital transformation of the built world. Our groundbreaking spatial computing platform turns buildings into data making every space more valuable and accessible. Millions of buildings in more than 170 countries have been transformed into immersive Matterport digital twins to improve every part of the building lifecycle from planning construction and operations to documentation appraisal and marketing.

About the Role:

Matterport has captured one of the largest and most diverse repositories of real-world 3D spatial data in existence. We are building a team to turn that data into a foundational reconstruction model a model that can reconstruct complete understand and generate photorealistic 3D spaces from sparse incomplete or noisy captures.

As a Senior Machine Learning R&D Engineer on this team you will research and build the next generation of neural reconstruction and generative systems that power Matterports digital twins. Rather than reconstructing only what the camera saw our goal is to generatively complete spaces filling occlusions synthesizing unseen views upgrading capture fidelity and ultimately generating entire coherent 3D environments. You will work at the intersection of neural rendering novel view synthesis and generative modeling training large-scale models on Matterports proprietary data to solve problems that no existing dataset or model can address.

This is a hands-on research-and-engineering role for someone who wants to push the state of the art in 3D generative AI and ship it into a product used across 170 countries. You will collaborate closely with fellow researchers platform engineers and product teams to translate frontier ideas into capabilities our customers can feel.

Responsibilities:

• Research design and train large-scale foundational models for 3D reconstruction and generative scene completion leveraging Matterports proprietary corpus of real-world spatial data.
• Advance the state of the art in neural rendering (e.g. Gaussian splatting NeRFs feed-forward/large reconstruction models) and novel view synthesis for real-world indoor and outdoor spaces.
• Develop generative approaches including diffusion models and video/3D generation to complete occluded or missing regions synthesize unobserved viewpoints and enhance the realism and completeness of digital twins.
• Design implement and optimize scalable training and inference pipelines for large 3D and multi-view datasets including images video point clouds and meshes.
• Translate research prototypes into production-ready systems partnering with engineering teams to integrate models into Matterports core capture and reconstruction platform.
• Define rigorous evaluation methodologies and benchmarks for reconstruction quality view synthesis fidelity and generative plausibility; drive iterative improvements against them.
• Stay current with advances across neural rendering generative modeling and 3D vision and proactively identify opportunities to apply new techniques to Matterports challenges.
• Collaborate with product managers software engineers and researchers to scope projects assess technical feasibility and deliver high-impact features.
• Contribute to the architectural design and strategic roadmap of Matterports foundational reconstruction model and broader ML research initiatives.
• Contribute to Matterports intellectual property through patents and publications and represent our work to the broader research community.

BasicQualifications:

• Bachelorsdegreefromanaccreditednot-for-profitin-personUniversityorCollege
• Atrackrecordofcommitmenttoprioremployers.
• Experience in machine learning research and development with a strong focus on computer vision 3D data processing neural rendering or generative modeling.
• Proficiency in PyTorch and strong programming skills in Python.
• Solid understanding of machine learning fundamentals including deep learning architectures (e.g. CNNs Transformers diffusion models) optimization techniques and evaluation methodologies.
• Experience with data manipulation and analysis libraries (e.g. NumPy Pandas).
• Demonstrated ability to conduct independent research propose novel solutions and implement them effectively.
• Excellent problem-solving skills and the ability to work with complex real-world datasets.
• Strong verbal and written communication skills

PreferredQualifications:

• Ph.D. or Masters Degree in Computer Science Computer Vision Machine Learning Graphics Robotics or a related quantitative field.
• 5 years of industry experience in applied ML research and development particularly training and shipping large models to production.
• Demonstrated expertise in building or training foundation / large-scale models and/or depth in two or more of the following:
  • Neural rendering & novel view synthesis Gaussian splatting NeRFs large/feed-forward reconstruction models differentiable rendering.
  • Generative modeling diffusion models image/video generation 3D generation world models.
  • 3D computer vision 3D reconstruction SLAM/structure-from-motion multi-view geometry point cloud and mesh processing depth estimation.

• Experience with large-scale model training infrastructure distributed/multi-GPU training large-dataset curation and pipelines and scaling to large model and data sizes.
• Strong track record of publications at top-tier venues (e.g. CVPR ICCV ECCV NeurIPS ICML SIGGRAPH) or significant open-source contributions.
• Ability to use AI tools to accelerate work.
• Experience mentoring junior engineers or leading technical projects.
• Ability to articulate complex technical concepts to a diverse audience.

Whatsinitforyou

WhenyoujoinCoStarGroupyoullexperienceacollaborativeandinnovativecultureworkingalongsidethebestandbrightesttoempowerourpeopleandcustomerstosucceed.

professionaland academic growth with internal training and tuition reimbursement.

• Ourbenefitspackageincludes(butisnotlimitedto):
• Comprehensivehealthcarecoverage:Medical/Vision/Dental/PrescriptionDrug
• Lifelegalandsupplementaryinsurance
• Virtualandinpersonmentalhealthcounselingservicesforindividualsandfamily
• Commuterandparkingbenefits
• 401(K)retirementplanwithmatchingcontributions
• Employeestockpurchaseplan
• Paidtimeoff
• Tuitionreimbursement
• Onsitefitnesscenterand/orreimbursedfitnesscentermembershipcosts(locationdependent)
• AccesstoCoStarGroupsEmployeeResourceGroups
• Complimentarygourmetcoffeeteahotchocolatefreshfruitandotherhealthysnacks

The final salary or hourly rate offered for this role will fall within the range set forth below based on a variety of factors including but not limited to geographic location skills and competencies.

BaseCompensation:$270000-$307000(Annually)

#Matterport

#LI-VP1

CoStar Group is an Equal Employment Opportunity Employer; we maintain a drug-free workplace and perform pre-employment substance abuse testing