AI Solution Architect (Generative AI)

Quick Facts

• Role: AI / Solution Architect for Generative AI (GenAI)

• Focus: end-to-end solution architecture, agentic workflows, RAG, and production deployment

• Work model: long-term project, remote; communication in English

Description

Translate business requirements into functional technical architecture and help deliver it as a working application. Design and implement agentic workflows in Python (LangGraph) with robust control (explicit state model, retry/timeout policies, human-in-the-loop checkpoints) and build a production-ready RAG layer (ingestion, embeddings, hybrid retrieval, domain isolation, and knowledge asset versioning). Specify tool contracts using MCP, clearly separate probabilistic vs deterministic logic, and design an Azure-based runtime with measurable and observable outputs.

Responsibilities

• Collect and refine business requirements with the client; clarify scope and estimate effort

• Convert requirements into technical design (architecture, integrations, costs, operations)

• Design and implement agentic workflow orchestration in Python using LangGraph

• Design RAG: ingestion, embedding, hybrid retrieval (vector + full-text), knowledge-domain isolation, knowledge-asset versioning

• Define MCP tool contracts (operations, schemas, idempotency, authorization, error handling) and integrate into agents

• Separate probabilistic parts (LLMs/agents) from deterministic validation and decision logic

• Design Azure runtime architecture and operation (Container Apps, Azure OpenAI, Document Intelligence, message bus, Key Vault, Entra ID)

• Define measurable output strategy (confidence indicators, escalation thresholds, evaluation sets, regression testing)

• Propose auditability and observability (OpenTelemetry, correlation IDs across the run, tracing prompt/model/config versions)

• Present and defend the design to developers (including client-side), conduct reviews, and maintain technical direction

• Make trade-offs between LLM-based behavior and classical automation

• Track GenAI progress and translate it into ongoing solution improvements

Requirements

• End-to-end experience designing and delivering AI/data solutions (not only partial implementation)

• Strong Python with ability to develop independently (FastAPI, Django, or Flask; REST API with OpenAPI)

• Practical experience with LLM applications: RAG, prompt engineering, agent orchestration (LangGraph/LangChain), and evaluation of output quality

• Practical understanding of Microsoft Azure architecture and infrastructure for deployment and operations

• Database knowledge and data model design (SQL, NoSQL, vector stores)

• Client-facing communication skills: gather real needs and explain technical designs to non-technical stakeholders

• English level for running client workshops (C1 or strong B2) and Czech for internal team collaboration

• Independence and willingness to take ownership of the results

Benefits

• Long-term project for a major Austrian insurance company

• Remote work; client communication in English and internal collaboration in Czech