Networking AI and Systems Lab
We study computer systems and AI applications. We focus on speed, scale, cost, and reliability.
Research in networking and AI systems.
All publications-
SceneHub4D
A dataset and evaluation framework for real-world 4D scenes.
IEEE TVCG · 2026 -
Reconstructing Reality over Time
Drone capture and timelapse Gaussian splatting for changing construction sites.
IEEE ISMAR Poster · 2025 -
BlenDR
Bandwidth-efficient RGB-D streaming for live 3D video.
ACM MobiSys · 2026 -
Neural-Enhanced Live Streaming
Online learning for better video quality during live ingest.
ACM SIGCOMM · 2020 -
Real-world 3D reconstruction
In VR, viewers can move through the room and look in any direction.
SceneHub4D · Demo -
Game streaming
Compare WebRTC and LiveNAS on the same League of Legends stream.
Live streaming · Demo -
Volumetric video capture
Cameras record color and depth from several angles at once.
SceneHub4D · Capture setup
Now at NAIS Lab.
Latest news.
-
FlaME accepted to USENIX NSDI 2027
-
NVIDIA supports our spatial intelligence platform with four workstation GPUs
-
NRF awards our lab an Outstanding Young Scientist Grant
-
BlenDR accepted to ACM MobiSys 2026
-
Neural-centric video processing accepted to CVPR 2026
-
SceneHub4D accepted to IEEE TVCG
-
NerVast accepted to WACV 2026
-
NAIS Lab launches its website
Ongoing research topics
- Efficient VLM
- Agentic AI systems
- AI-native mobile networks
- Immersive media and XR
- LLM applications and systems