🌌 I'm Jiahao Zheng · aka "++"

Undergraduate Researcher at Nanjing University of Posts and Telecommunications (NJUPT)
📍 Majoring in Computer Science and Artificial Intelligence

I am an AI researcher dedicated to bridging the gap between Computer Science and Natural Sciences. My work currently focuses on:

• 🦾 Embodied AI & Robotics - Building agents that understand and manipulate the physical world
• 🧠 Transformer Architectures - Leveraging attention mechanisms for multimodal understanding
• 🎯 Action Chunking Transformers (ACT) - Pioneering imitation learning for robotic manipulation
• 🔬 AI for Science (AI4S) - Accelerating scientific discovery through intelligent computation
• 🤖 Multimodal Agents - Integrating vision, language, and action for intelligent systems

💡 It's a clever pun on my Chinese name:

• 🔤 Jiahao sounds like "Plus Sign" (加号)
• ➕ Zheng means "Positive" (正)

So mathematically, I am literally ++ 😄
Like the increment operator - always improving, always growing!

🤖 Embodied Intelligence Core Algorithms: 🎯 ACT (Action Chunking Transformer) Temporal action sequence modeling Imitation learning for robotic manipulation CVAE-based policy learning 🦾 Vision-Language-Action Models Cross-modal fusion architectures End-to-end perception-action pipelines Real-world robotic deployment

🧠 Transformer Architectures Technical Depth: 🔍 Self-Attention Mechanisms Multi-head attention optimization Positional encoding strategies Sparse attention patterns 📊 Multimodal Transformers Vision-Language pre-training Cross-modal alignment techniques Efficient fine-tuning methods (LoRA, Adapter)

I am actively expanding my research horizons into the following frontier areas:

🤖 AI & Deep Learning Frameworks

🦾 Robotics & Embodied AI

💻 Development & Research Tools

🔬 Scientific Computing

🎨 Productivity & Documentation

🎮 Mobile & Others

My engineering practice spans diverse domains, focusing on Embodied AI and Scientific Computing. I have developed intelligent agent systems and utilized VLA (Vision-Language-Action) models for real-world applications.

🐍 GitHub Contribution Snake

I am an avid marathon runner, believing that the endurance required on the track mirrors the persistence needed in scientific research.

"Every kilometer on the track is a metaphor for every epoch in training neural networks."

What Running Teaches Me:

• Patience - Models converge slowly, just like endurance builds gradually
• Resilience - Hitting the wall at km 30 ≈ NaN loss at epoch 87
• Optimization - Finding the right pace ≈ Finding the right learning rate

My idols, LeBron James and Kylian Mbappé, have profoundly shaped my understanding of dedication and peak performance.

LeBron James The NBA's all-time leading scorer and a defining figure of athletic longevity and leadership.

Kylian Mbappé FIFA World Cup winner and one of the most explosive and talented forwards in modern football.

💭 Random Dev Quote

⭐️ From jiahaozheng406 · "++" · The Incremental Optimist

_{"Coding for AI, Coding for Science. Exploring the Physics of Intelligence." 🌌}