Hi, I'm Hongjing Wang, a software engineering student at Dalian University of Technology.

I'm drawn to one ambitious question: how do we build intelligence that can act in the real world, carry out sustained work, and improve the way it learns?

• 🤖 Embodied intelligence — exploring the loop from perception and world models to planning, action, and feedback. Intelligence gets interesting when it has to meet reality.
• 🛠️ Agent harnesses — interested in the infrastructure that turns model capability into dependable execution: tools, memory, context, orchestration, and evaluation.
• 🔁 RSI — fascinated by recursive self-improvement: systems that can inspect their failures, refine their methods, and earn their next capability through measurable progress.
• 🚀 I enjoy turning ambitious ideas into working systems, tracing problems across the stack, and pushing prototypes until the hard questions become concrete.

The ambition: build systems that act with purpose, learn from experience, and compound their capabilities.

My north star is the intersection: agents that learn by doing, harnesses that make execution reliable, and feedback loops that turn experience into improvement.

Think from first principles. Build the whole loop. Let experiments settle the argument.

I'm especially interested in thoughtful engineering, ambitious research ideas, and open-source work that makes advanced AI easier to understand and build on. If you're exploring embodied agents, agent infrastructure, or self-improving systems, I'd love to connect through GitHub.

Embodied intelligence enables intelligence to enter the real world. Harness turns capabilities into reliable execution. RSI makes every exploration the starting point for the next evolution.

_{Curiosity sets the direction. Engineering makes it real. Feedback moves it forward.}