Researcher focusing on LLM infrastructure efficiency and hyperscale observability.

Developing SDAG (Systematic Defect Awareness & Guidance) to bridge the gap between model-level routing decisions and physical hardware reliability. Interested in MoE stability, tail latency analysis, and semiconductor degradation modeling at scale.

GitHub: https://github.com/alexbuiko-sketch Open for deep technical discussions on AI infra.