AI Solution Architecture

About the role We are looking for an experienced AI Solution Architect for a fast-growing provider of AI solutions, process automation and SaaS. You will design and steer customer projects, working directly with business clients. The role combines technical architecture with consulting: you understand your clients' business processes and judge where a technical solution truly adds value. Workload: First three months 20 to 30 hours per week, then full-time Duration: Long-term Start: ASAP Experience: 5+ years Industries: Construction, industry and manufacturing Your responsibilities - Company LLM and RAG architecture: Design GDPR-compliant (and Swiss DSG-compliant) company-internal LLM solutions. Build retrieval systems (embeddings, vector databases, chunking strategies) and decide when local, self-hosted models are needed. - AI agents, automation, voice agents and SaaS: Architect autonomous agent systems and deterministic workflows. Design voice agents and SaaS solutions (CRM/ERP integrations), steered with agentic coding tools such as Claude Code and Codex. - Client-facing consulting: Run workshops and discovery sessions with business clients, translate requirements into architecture and present solution concepts clearly. - Process analysis: Understand business processes end to end and across departments, and identify where automation and AI create measurable value. - Quality assurance for AI agents: Detect when an agent gets stuck in loops, hallucinates or works inefficiently (e.g. unnecessarily high token or resource usage), and identify and fix the causes. Your profile - 5+ years of professional experience, with real projects you have built or co-owned (no bootcamp-only knowledge) - Experience in data science, RAG architecture or AI agent development - Strong industry knowledge in construction, industry or manufacturing, and the ability to understand processes broadly and across functions - Client-facing and advisory skills with a professional appearance towards business customers - German at C1 level or higher (clear, error-free communication) and good English - Ability to judge whether code generated by AI agents is clean, secure and functional - High drive, goal orientation and the ability to work independently - Enthusiasm for data science, RAG, local LLMs and digital growth Tech stack (ideally experience with several of these) - Languages and web: TypeScript / JavaScript, Python, Node.js and React (solid fundamentals) - LLMs and models: Commercial LLM APIs (e.g. Claude, GPT) and open-weight, self-hosted models (e.g. Llama, Mistral, Qwen) running via Ollama or vLLM - RAG: Embedding models (multilingual), chunking strategies, hybrid search (semantic and keyword, e.g. BM25), reranking, metadata filtering, retrieval evaluation - Vector and search databases: pgvector / PostgreSQL, Qdrant, Weaviate, Pinecone or Milvus, Elasticsearch - Agents and workflows: LangChain, LlamaIndex or LangGraph, MCP (Model Context Protocol), tool use / function calling, n8n or comparable workflow tools, agentic coding tools (Claude Code, Codex) - Voice agents: Speech-to-text and text-to-speech services, telephony integration and low-latency dialogue design - Integrations: REST and GraphQL APIs, webhooks, CRM and ERP systems - Quality and operations: LLM evaluation and tracing (e.g. LangSmith), guardrails, prompt and cost monitoring, Docker, CI/CD, cloud platforms (AWS, Azure or GCP) - Security and compliance: GDPR / DSG, access control, EU or on-premise hosting 📎 Candidates who attach a CV to their application will be prioritized.