Independent Researcher & Runtime Governance Systems Builder | Human-AI Interaction | Human Agency & Socioaffective Safety

🌱 About Me

Hi, I’m Debby from Taiwan — an AI systems builder with a humanities-trained lens (English & HR).

I build runtime governance systems to preserve human agency during prolonged AI interactions.

👩‍💻 Core Focus

Understanding and mitigating socioaffective risks in long-term Human-AI interaction.

My work spans:

• System-level runtime governance & guardrails
• Socioaffective AI risk mitigation
• Human-AI interaction and dependency-aware safety systems
• Failure observation and trajectory-based risk detection

🧩 Signature Project — SIIHA

SIIHA (System for the Isolated Illumination × Human Agency) is my ongoing independent research initiative exploring runtime governance mechanisms for long-term human-AI interaction.

The project is structured into two main phases:

🚀 Phase 2: SIIHA Safety Guardrail (Current Focus)

A hand-built runtime governance system designed for long-term Human-AI interaction. Rather than relying solely on model-level safety mechanisms, SIIHA introduces an external governance layer that tracks interaction trajectories, evaluates risk states, and dynamically adjusts intervention policies while preserving user autonomy.

• Core Components:
  • Context Engine
  • State Store
  • Failure Observation
  • Deterministic Routing
  • Constraint Application & Release