AI & ML interests
Physical AI, Agentic AI, on-device and edge intelligence, multimodal perception, context and state modeling, tool use and capability routing, small language models, VLM/VLA, agent-device interaction, and efficient inference for intelligent devices.
Recent Activity
Identity · Capability · Proximity · Coordination · Memory · Evolution
INT AIOS
Infrastructure for the Physical Agent Era.
INT AIOS is building an AI-native runtime and infrastructure layer for agents, devices, and physical spaces.
We believe the next generation of AI will move beyond chat interfaces and become part of the physical world — continuously perceiving context, understanding state, discovering capabilities, and acting across devices.
Our work focuses on:
- Agent Runtime — execution and coordination infrastructure for persistent agents.
- Device Agents — turning physical devices into discoverable and callable agent capabilities.
- Context & State — building persistent, structured representations of people, devices, and spaces.
- Capability Protocols — standardized schemas for discovering, describing, invoking, and composing capabilities.
- Edge & On-device AI — efficient models for perception, intent understanding, capability routing, and local execution.
- Agent Internet — infrastructure for trusted interaction between agents, devices, services, and physical environments.
Our long-term goal is to build the runtime layer connecting:
Perception → Context → Reasoning → Capability → Action
We are exploring and open-sourcing models, datasets, protocols, runtimes, and tools for the emerging Physical Agent ecosystem.
Device is Agent. Context is Product. Capability is the Interface.