I build intelligent systems, and the guardrails that keep them from going rogue.
AI developer focused on LLM integration, agentic systems, and the security layer between AI models and the real world. Currently building production-grade tooling to make autonomous agents safer, more auditable, and harder to exploit.
400-hour Master in AI Development · ProfessionAI · 28/30, covering LLMs, Agentic AI, REST APIs and MLOps.
Security middleware for AI Agents. Multi-layer defense: Binary Allowlisting → Regex Heuristics → LLM Intent Coherence. Cryptographically signed logs via HMAC-SHA256 for forensic integrity.
Input ──► L1: Allowlist ──► L2: Pattern Engine ──► L3: LLM Intent Check ──► Agent
│ │ │
BLOCK BLOCK BLOCK + log
AI & Security
Backend & Infrastructure
Automation
- AI Agent Safety: stopping autonomous agents from hallucinating destructive shell commands
- LLM Pipelines: building secure, observable inference pipelines with self-hosted models
- Workflow Automation: n8n + Docker + local LLMs for real-world automation
- Maker projects: merging AI with 3D printing for smart hardware prototypes
When I'm not pushing commits: riding mountain routes across the Alps on two wheels, debugging long 3D prints, and deep-diving into motorsport data.