05.07.2026 aktualisiert
100 % verfügbar
AI Agent & RAG Engineer | Python, FastAPI, LangGraph | AI Governance
viganello, Schweiz
Weltweit
IBM RAG & Agentic AI; IBM Generative AI Engineering; security/access control background.Über mich
Independent AI Engineer based in Lugano. I build private RAG assistants, agentic workflows and AI automation with clear scope, audit logs, permission boundaries and human-in-the-loop approval.
Skills
AI AgentsRAGLLMPythonPostgreSQLMachine LearningRedisGenerative AIFastAPIDocker
I design and implement governed AI systems for Swiss SMEs, agencies and technical teams.
Core focus:
- Private RAG assistants over company documents, policies, manuals and knowledge bases
- Agentic AI workflows with human-in-the-loop approval, audit logs and permission boundaries
- AI workflow automation for operations, research, sales and support
- Architecture reviews before investing in AI MVPs or production systems
Technical stack:
Python, FastAPI, LangGraph, PostgreSQL/pgvector, Redis, Docker, REST APIs, WebSockets, Claude, OpenAI-compatible APIs, AWS Bedrock-style architectures, Ollama/local model workflows, vector search and backend integration.
What I bring:
- Practical AI delivery, not just demos
- Clear scope, measurable pilot outcomes and realistic next steps
- Governance-oriented and privacy-aware implementation
- Experience with security-sensitive systems, access control, reliability and technical documentation
Best-fit mandates:
- RAG / LLM Engineer
- AI Agent Engineer
- Python / FastAPI backend for AI applications
- AI automation consultant
- AI governance / architecture review
- Prototype-to-MVP support for founders, agencies and IT providers
Based in Lugano, available remote across Switzerland and on-site when useful. Minimum rate target: CHF 120/h; preferred range CHF 130-150/h or project/day rate for governance and architecture work.
Erfahrung
Berufserfahrung5 - 10 Jahre
Freelancer seit2026
Sprachen
DeutschGrundkenntnisseEnglischgutItalienischMuttersprache
Projekthistorie
Private AI engineering portfolio documenting case studies from private repositories without exposing code, credentials or client data.
Built and documented:
- MirrorSkills / Governed Enterprise Agents: enterprise agentic AI platform concept with tenant isolation, permission controls, Doer/Critic review, human approval and audit-oriented records.
- AI & ML Agent: FastAPI/LangGraph runtime with REST/WebSocket interfaces, controlled tool execution, hybrid RAG over PostgreSQL/pgvector, Redis services and provider fallback.
- VisibilAI: GEO/AEO analytics concept for measuring answer-engine visibility, structured data, llms.txt generation and action plans.
- Realtime Translator: local Windows MVP using faster-whisper, NLLB/CTranslate2, WebSocket streaming, voice-send gate, consent/audit diagnostics.
- RedBenchOps / Syntrocyber: defensive AI security work around prompt injection, tool allowlists, memory boundaries, testing campaigns and responsible reporting.
Stack: Python, FastAPI, LangGraph, PostgreSQL/pgvector, Redis, Docker, TypeScript/Next.js, React, Playwright, Claude/OpenAI-compatible APIs, AWS Bedrock-style architectures, Ollama/local models.
Public summary: github.com/giack891811/ai-engineering-portfolio