At Compass, we envision a world where the experience of selling or buying a home is simple and pleasant for everyone. Founded in 2012, Compass provides an end-to-end platform that empowers residential real estate agents to deliver exceptional service to their seller and buyer clients, all in service of our mission to help everyone find their place in the world.
About The Team
The Compass Computer Vision (Video and RE/COgnition) team makes it easy for customers to add text, image and video analysis to real estate applications with no machine learning expertise required. With Compass Computer Vision, you can enrich listing descriptions, tag rooms and listings, identify objects and scenes in text, image and video providing immersive virtual tour experiences from physical or creative data sources.
Our team's goal is to disruptively improve customers' experience and business by creating a text, image and video-analysis platform for real estate applications. This includes:
Developing text, image and video-analysis solutions for agents to leverage to grow their business. Agents and their clients can easily understand the problems we are solving, and will share our conviction that these solutions will help them win and serve more clients.
Embracing radical simplicity. We strive to deliver very simple user experiences that enable agents to adopt with as little effort as possible. We also prefer simple and scalable solutions to complex ones.
Not reinventing the wheel. We take pains to benchmark and understand state of the art open source and cloud AI solutions. We use this learning to achieve speed and quality in our work, and leverage existing tools when it makes sense to #MoveFast on behalf of our customers.
Doing rapid prototyping to test ideas with customers. We believe that low cost mockups, hackathons and quick prototypes are invaluable for learning what works. We are not afraid of failed prototypes or ideas that didn’t work when tested with customers, because we believe that even a few amazing wins from our process of rapid iteration will more than compensate for early failures.
Building APIs that are productized, platformized and reusable. We build and own well-designed APIs that can be easily integrated by many Compass applications, and follow the best practices of API design, documentation and support.
About The Role
As a Senior Machine Learning Scientist on the Computer Vision team, you will work closely with engineers, designers, and product managers to invent software prototypes for image classification, object detection, video understanding, rendering, 3D reconstruction technologies in the real estate industry. You will have the opportunity to help create, build, deploy and test novel image classification, object detection, video understanding, 3D reconstruction, image to caption systems and algorithms. You will have an impact on shipping applications with millions of users and help define this new effort from the start. You will also have an opportunity to create/lead image classification, object detection, video understanding, rendering, and 3D reconstruction projects
At Compass You Will:
Build, develop and deploy performant and scalable image processing/enhancement, video stitching services
Build, develop and deploy image classification, object detection, video understanding, 3D reconstruction, image to caption services
Collaborate with product managers and work with an engineering-focused, iterative team to build and establish product requirements
Quickly prototype new demos and systems for real estate image and video understanding
Quickly prototype new demos and systems for image classification, object detection, video understanding, rendering, 3D reconstruction, image to caption
Build systems on top of state-of-the-art CV algorithms across Compass’ platforms, as well as third-party services
Iterate and prototype rapidly
What We Look For:
MS in Computer Vision, Computer Graphics, or Computer Science.
5+ years of relevant experience
Must possess a strong background in Computer Vision or Computer Graphics
Experience with machine learning algorithms such as CNN, Resnet, Pytorch, Tensorflow, etc.
Experience in optimization on GPU / CPU / other architectures (CUDA, SSE, NEON, OpenMP or other SIMD)
Experience with deep learning / object reconstruction / registration / classification / recognition / rendering
Testing, documentation, and verification of or image classification, object detection, video understanding, rendering, 3D reconstruction, image to caption
Experience with Agile methods, Scrum / Kanban / etc.
Knowledge of scripting language, e.g. Python, Bash, etc.
Camera and/or image/or graphics/or video pipeline knowledge and experience
Nice to Have:
Ph.D. in Computer Vision, Computer Graphics, or Computer Science.
Prior experience involving image processing, computer graphics, video processing, geometrical computer vision is a big plus.
Prior experience involving image classification, object detection, video understanding, rendering, 3D reconstruction, image to caption is a big plus.
Experience with existing computer vision libraries such as OpenCV, PCL, CGAL, Eigen, etc.