I explore the boundary between AI systems, compilers, and hardware-aware optimization.

I don't just use models — I interrogate them.

• 🔧 AI Compiler & Kernel-level Optimization (PyTorch / Triton)
• 🧠 Failure-aware Systems & Dynamic Shape Handling
• ⚙️ Hardware-conscious Model Execution (GPU / NPU)

• Failure is not noise — it is signal
• Optimization without understanding is just luck
• Systems should adapt, not just converge

🌀 Phase I — Snake

🐉 Phase II — Dragon