Spatial AI Research for AR/VR

Emerging AR/VR and wearable devices require AI systems that understand interaction, spatial context, and physical environments. The technical challenge is not just model performance; it is translating early-stage research into working device-oriented prototypes.

NDA-sensitive research environment

Device-oriented prototype constraints

Spatial and scene-identification ambiguity

Multimodal representation challenges

Need to turn research concepts into demonstrable systems

Worked on AI-driven capabilities for emerging AR/VR and wearable devices, including spatial-AI research involving 3D scene representations, multimodal embeddings, Gaussian splatting, surfels, and CLIP-like embedding models.

Also owned software implementation for a hand-control interaction demo on wearable hardware.

Built and evaluated working research prototypes and produced evidence for potential improvements in spatial and scene-identification workflows.

Some outcomes and implementation details are intentionally generalized because of NDA constraints.

This case is relevant to teams working on spatial AI, robotics, AR/VR, 3D scene understanding, multimodal embeddings, device prototypes, or research-to-product execution.

Details are summarized at a high level due to NDA-sensitive work.

Working demo

device-oriented prototype NDA-sensitive

research context Spatial AI

Python, Embeddings, CLIP-like models, Gaussian splatting, Surfels, AR/VR prototypes