I work at the intersection of systems design, security analysis, and real-world deployment risk.
→ Start Here — Who I am, how I work, and why.
Identifying structural failure modes in language-based AI systems, especially those that emerge during real interaction rather than scripted evaluation.
- Premature intent resolution and confidence laundering
- Pre-articulation observability limits
- Probabilistic identity and contestability gaps
- Deployment risk in child-facing and safety-critical contexts
| Repository | Description |
|---|---|
| llm-epistemic-honesty | They Lie, and We Lie About the Lying — Why "hallucination" is a euphemism. 35 references. Policy recommendations. |
| symbol-grounding-critique | On the Inability of Language Models to Stub Their Toe — Referential collapse and semantic inflation |
| Repository | Description |
|---|---|
| start-here | Full methodology and background |
| pre-articulation-boundary | Position paper on structural limits of language-based AI |
| uivp-disclosure-summary | Cross-model intent misclassification disclosure |
| probabilistic-identity-security-analysis | Security analysis of emergent identity infrastructure |
| revshare-forensic-case-study | Forensic analysis of custodial architecture failure |
Before focusing on independent research, I spent years designing and operating real systems in production environments:
- Led operational teams in safety-critical infrastructure
- Built systems handling millions of daily users
- Designed custody models, incentive systems, and abuse-resistant architectures
- Performed post-incident forensic analysis when systems failed
- Exploratory first, disciplined second
- Anomaly sensing before formal diagnosis
- Documentation and reproducibility over hype
- Systems evaluated as deployed artifacts, not just models
Website: cpk.solutions
Email: christopher@cpk.solutions
I am open to responsible disclosure conversations, research collaboration, and advisory work aligned with safety, integrity, and system design.