Senior Computer Vision Engineer, Embodied AI

About Niantic Spatial

AI has learned from almost everything online, but it still struggles with the physical world, where 99.5% of economic activity happens. It doesn’t know where things are, how spaces fit together, or what changes over time.

That’s one of the biggest unsolved problems in AI, and it’s the one we’re solving at Niantic Spatial. We build spatial AI models and services so people and machines can reason, learn and act safely in the physical world. We turn camera imagery, video, sensor data and archives into accurate, AI-readable representations of real places that robots can train in and machines can use to know exactly where they are.

In physical AI, mistakes have real consequences. That’s why we build AI grounded in measured reality, not guesswork, so the people and machines relying on it can trust it to get things right.

Our team built some of the world’s most widely used maps. Now we’re building something bigger: a living, measured record of the real world that machines can query, learn from and use to understand how places change over time. If you want to help AI understand the world we actually live in, come build with us.

About the Role

Niantic Spatial makes the physical world computable, helping people and machines collaborate safely by aligning how they understand reality. One of the fundamental problems we're helping autonomy teams and engineers overcome is the sim-to-real gap for visual-spatial understanding. We're applying our team's decades of experience encoding the world precisely as it is at Google Maps, Google Earth, and Niantic Labs to build the scalable real-to-sim stack for embodied AI.

We're looking for a Senior Computer Vision Engineer to join our Embodied AI team, focused on 3D scene understanding and semantics. A reconstructed environment is only useful to a robot once it knows what is in it: which surfaces are floor, which objects can be grasped, moved, or opened, where one room ends and another begins. You'll work hand in hand with our Research Scientists to take advances in open-vocabulary perception, 3D semantic labeling, and scene graph construction and turn them into reliable capabilities that embodied AI teams can use in production.

This is an applied research role. You'll live at the boundary between a research prototype and a production pipeline: adapting methods so they hold up on messy customer captures, lifting 2D semantic signals into geometrically consistent 3D, defining what "semantically correct" means for a downstream policy, and feeding what you learn back into the research agenda. You understand that the difficult part is often not the core method, but the messy inputs, edge cases, operational constraints, and downstream requirements surrounding it. You care about quality, cost, latency, and reliability, and you know that a system is only successful when people can depend on it. This is a hands-on engineering role for someone who wants to build, ship, and own outcomes. You'll help shape both the technology and the product, while remaining close to the code and the problems our customers are trying to solve.

What You'll Do

Productionize 3D Scene Understanding Capabilities - Turn research prototypes in open-vocabulary segmentation, 3D semantic and instance labeling, affordance prediction, and scene graph construction into reliable pipeline components and services that operate across varied customer data and environments.

Own End-to-End Output Quality - Ensure that reconstructed environments carry labels, instances, and relationships that are consistent, complete, and correct enough for downstream simulation and policy learning, and define what "correct enough" means for each customer workflow.

Bridge Research and Production - Work day to day with the Embodied AI Research Scientist: pressure-test new methods on real customer data, identify where they break, and bring production and customer constraints back into research planning to prioritize the advances that matter most.

Make Systems Robust to the Real World - Diagnose failures caused by capture quality, scene complexity, scale, calibration, coordinate conventions, and other assumptions that prototypes often leave implicit.

Build Evaluation and Quality Infrastructure - Create datasets, regression tests, quality gates, and benchmarking tools that help the team measure whether changes improve the system.

Improve Performance, Throughput, and Cost - Profile and optimize GPU and distributed workloads, reduce unnecessary reruns, and help establish the economics of running reconstruction at scale.

Work Across the Product Boundary - Partner with Embodied AI product and engineering teams to understand customer requirements and deliver capabilities that fit real training, evaluation, and deployment workflows. Help evolve the representations, tooling, and operational systems that allow reconstructed environments to move reliably into customer applications.

What You'll Bring

5+ years of computer vision experience and a bachelor's degree or equivalent,

Significant experience building and operating production computer-vision, machine-learning, graphics, or data-processing systems.

Deep hands-on experience with scene understanding: semantic, instance, or panoptic segmentation; open-vocabulary detection and segmentation; or vision-language models applied to perception.

Experience taking 2D perception into 3D, such as multi-view label fusion, 3D semantic segmentation, instance tracking across views, or scene graph construction over reconstructed geometry.

Experience delivering reliable computer-vision or 3D systems used by other teams or customers, including the evaluation that proved they worked..

Practical command of 3D and spatial data, including geometry, camera models, coordinate frames, calibration, and metric scale.

Strong Python and PyTorch skills, with the ability to work in C++ or other performance-oriented environments when needed.

Experience with cloud infrastructure, GPU workloads, distributed processing, or large-scale data pipelines.

Nice to Have

Hands-on experience with 3D reconstruction, photogrammetry, structure from motion, Gaussian splatting, neural rendering, or meshing.

Experience with SLAM, visual positioning, localization, or large-scale mapping systems.

Experience with robotics, simulation, or embodied AI applications, in particular how semantic labels are consumed by a policy or a simulator.

Built evaluation or benchmarking infrastructure for perception or reconstruction systems.

Experience optimizing large-scale GPU inference workloads for cost, throughput, or latency.

A publication record or contributions to widely used open-source perception codebases.

Advanced degree in computer vision, robotics, machine learning, or a related field.

Competencies

Relentless bias for action. You make decisions and ship as if the company’s success depends on it, because it does. You set the pace, and drive the people around you to rise to it.

Intellectually honest. You know there’s just one job that underpins every other in a startup: find the truth. You are relentless in your pursuit of it, including when it challenges your own convictions, and you raise concerns when you see them — e