Developer focused on structure, audit, and operational clarity.
Working across AI / LLM evaluation, decision systems, automation, and black-box handling.
The goal is not only to make systems work, but to make them bounded, inspectable, and deployable.
| Platform | Domain | Focus |
|---|---|---|
| blue-tanuki-core | AI / Control Core | Policy-driven execution, bounded routing, audit, and safe control |
| hds-public-shell | Decision Interface | Public shell for HDS-style routing with suspend/assert boundaries |
| GitHub Profile | Engineering Identity | Repositories, design direction, and public-facing implementation notes |
- AI / LLM Evaluation
- Audit-Oriented System Design
- Decision Routing
- Black-Box Handling
- Practical Automation
- Operational Safety
- Structure over noise
- Reproducibility over impression
- Explicit boundaries over ambiguity
- Safety before expansion
- Operational value over empty sophistication
I am currently focused on:
- building auditable AI-facing system components
- improving evaluation and benchmark design
- designing bounded execution and routing layers
- treating black-box systems as operational targets, not objects of blind trust
| Axis | Direction |
|---|---|
| Engineering Style | Implementation-focused |
| System Preference | Bounded and inspectable |
| Evaluation View | Audit before trust |
| Design Priority | Real-world operability |
| Risk Stance | Safety before speed |
- I prefer systems that can be tested, inspected, and recovered
- I treat ambiguity as something to bound, not decorate
- I value implementation that survives contact with reality
- GitHub: gatchimuchio
- Core Repo: blue-tanuki-core
- Public Shell: hds-public-shell