Research Scientist Intern, Low-Power Scene Reconstruction for AR/VR

Meta•Redmond, WA

7d•$7,650 - $12,667

About The Position

Reality Labs, Meta’s innovation engine for next-generation AR/VR and wearable technology, is developing an imaging and reconstruction stack for future glasses-like devices. This research internship focuses on low-power, real-time reconstruction of scene geometry for AR/VR, addressing the challenge of delivering high-quality passthrough rendering and scene understanding within tight power budgets. The work will investigate depth estimation and processing strategies designed for glasses, and whether a lightweight learned scene representation can support passthrough, hand occlusions, and background occlusions simultaneously. The intern will own an open research problem, choose methods, build an evaluation, and present results, with support for publication when warranted. The research spans depth estimation, occlusion, and geometric scene understanding, with evaluation on real multi-sensor device captures and collaboration across passthrough, scene understanding, and on-device compute teams.

Requirements

Currently has, or is in the process of obtaining, a PhD in computer vision, machine learning, computer graphics, robotics, or a related technical field.
Must obtain work authorization in the country of employment by the time of hire and maintain ongoing work authorization during employment.
Knowledge and hands-on experience in at least one of depth or stereo estimation; 3D reconstruction (TSDF, implicit or neural fields, Gaussian splatting); video or instance segmentation; or neural scene representations.
Experience designing and running quantitative experiments and communicating technical results.

Nice To Haves

Intent to return to a degree program after completing the internship.
Track record of significant results demonstrated by grants, fellowships, patents, or first-authored publications at leading workshops or conferences such as CVPR/ECCV/ICCV, SIGGRAPH/SIGGRAPH Asia, 3DV, ICLR, or NeurIPS.
Experience with efficient or on-device inference, including quantization, distillation, pruning, or deployment to mobile NPU, DSP, or GPU targets.
Experience with multi-sensor fusion across heterogeneous sensors, including RGB, IR/CV cameras, and time-of-flight, as well as camera calibration and cross-sensor registration.
Experience with temporal filtering and state estimation over scene state.
Familiarity with AR/VR passthrough, reprojection, occlusion rendering, or SLAM and scene reconstruction.
Experience with C++ and profiling or optimizing real-time computer vision or graphics pipelines.
Experience working in a Unix environment.
Experience reasoning about compute and power budgets as a first-class design constraint.

Responsibilities

Plan and execute research to advance low-power, real-time scene reconstruction for AR/VR, spanning depth estimation, occlusion, and geometric scene understanding.
Explore whether a shared learned scene representation can serve multiple downstream consumers with different latency and accuracy needs.
Define evaluation methodology and run experiments measuring geometric accuracy, temporal stability, rendering distortion, latency, and compute and power cost on real multi-sensor device captures.
Collaborate with researchers and engineers across passthrough, scene understanding, and on-device compute teams to assess feasibility within headset and glasses power envelopes.
Document and present findings, and contribute to publications and patents where results warrant it.