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

