AI Agent Specialist · AI Security — Make your agents wAIser.
$ whoami
> I build autonomous agents — and break them before attackers do.
$ current_target
> Memory poisoning in agent systems. How few messages does it take?- 🤖 AI Agents — autonomous agents, workflows & self-hosted LLMs that do real work. CrewAI · LangChain · RAG · Ollama
- 🛡️ AI Security — red-teaming & hardening for AI systems, from prompt-injection defense to cloud. 2nd place in a prompt-injection challenge.
- ⚙️ Custom Development — selective, by invitation. React Native · SvelteKit · Supabase · TypeScript
agent-poisoning— how many conversational messages does it take to poison a memory-backed AI agent? Mem0 Cloud vs lean-memory.lean-memory— embedded, local-first agent memory. No server, no daemon, no mandatory cloud key.
Raw notes on AI agents, AI security and the machines learning to move without us — published before the trail fades. → waiser.dev/blog (DE · EN · 中文)
📍 Lake Constance, Switzerland · 💬 waiser.dev — let's talk




