At Niantic Spatial, we’re building the future of geospatial AI. Powered by a proprietary database of over 30 billion posed images and a groundbreaking third-generation digital map, our mission is to develop spatial intelligence that helps both humans and machines better understand, navigate, and engage with the physical world. Our high-fidelity mapping technology unlocks a new dimension of interaction—laying the foundation for AI to truly comprehend and operate within real-world environments. Join us as we build a living model of the world that people and machines can talk to.
As a Computer Vision Engineer on the Applied Algorithms team, you will own significant parts of our Localize capabilities. While our broader engineering org builds the foundation of the Large Geospatial Model (LGM), this team is responsible for the our Visual Positioning System (VPS) that allows for centimeter-level precision in mapped areas and answers the question "where am I" for robots and humans and their devices alike. This information is an important building block to our services and LGM work. Bridging the gap between R&D and production, you will turn complex geospatial data into a persistent sense of space, enabling the next generation of spatial AI.
System Architecture: Design, develop, and maintain production-grade computer vision systems that power Niantic Spatial’s VPS (Visual Positioning System) and 3D mapping pipelines.
Algorithmic Innovation: Refine and scale algorithms for our VPS system including, SfM (Structure from Motion), efficient feature and descriptor extraction to have the best possible accuracy and recall..Moving algorithms from research concepts to hardened production code that can scale for our map of the world.
Performance Engineering: Optimize complex ML and CV code for maximum efficiency in cloud and mobile environments, ensuring low latency and high-performance execution on GPU/CPU.
Benchmarking & Evaluation: Create and own the tools and frameworks used to evaluate the quality of our spatial grounding and 3D maps against ground-truth data.
Technical Leadership: Lead technical design reviews, mentor junior engineers, and serve as a team anchor for resolving complex technical disagreements within the mapping stack.
Cross-Functional Delivery: Collaborate with Product, Research, and Operations teams to ensure that state-of-the-art computer vision solutions translate into delightful user experiences.
Education: BS, MS, or PhD in Computer Science, Robotics, Computer Vision, or a related technical field (or equivalent professional experience).
Years of Experience: 5+ years of experience developing and shipping algorithms for image processing, computer vision, or 3D reconstruction.
Coding Proficiency: Expert-level proficiency in python and/or C++.
Domain Expertise: Proven track record in designing solutions for Structure from Motion (SfM), VPS or 3D mapping..
Frameworks & Tools: Deep experience with Deep Learning frameworks (PyTorch or JAX) and version control (Git).
Required in-office days: 3 days per week
Plus If:
Experience in planning and leading technical projects from inception to production.
Significant contributions to open-source CV libraries (OpenCV, COLMAP, etc.).
Experience with CUDA or shader programming for performance optimization.