AI Engineer – Agentic Systems

PortBlueSky is a company building an internal analytical platform for a major multinational enterprise that combines a modern web application and enterprise data layer with AI agents to help users prepare and analyse data, generate visualisations, extract insights, and synthesise outputs.

• Strong Python development skills.
• Hands-on experience building LLM-powered applications or agentic systems beyond simple prompt/API integrations.
• Experience with orchestration frameworks such as LangChain, LangGraph, Google ADK, or comparable tooling.
• Strong understanding of agent orchestration, tool calling, state management, structured inputs and outputs, and failure handling.
• Practical experience with logging, tracing, observability, and debugging for AI applications.
• Understanding of evaluating and regression-testing agentic systems as prompts, models, tools, and workflows evolve.
• Awareness of production AI concerns including security, permissions, prompt injection, token usage, latency, reliability, and cost.
• Good software engineering practices including version control, pull requests, automated tests, CI/CD, containerisation, and clear application architecture.
• Ability to work independently amidst evolving requirements.
• Pragmatic attitude toward technology frameworks prioritizing solving problems effectively.
• Excellent communication skills and professional fluency in English.
• EU residency and permission to work in the EU.

Nice to have:

• Experience with FastAPI and production Python web applications.
• Experience with PostgreSQL.
• Experience with Docker and Docker Compose.
• Experience with AWS or other major cloud providers.
• Experience with Microsoft Fabric, Databricks, Snowflake, or comparable data platforms.
• Experience with pandas, Polars, SQL, or analytical data processing.
• Experience with Pydantic or other structured validation approaches.
• Experience with OpenTelemetry or comparable observability tooling.
• Experience with enterprise analytical or diagnostic applications.
• Experience with multi-agent architectures and building AI features on large structured datasets.

• Design agentic workflows involving agents and multi-step AI workflows for analysis, data preparation, visualisation, insight extraction, and synthesis.
• Own orchestration by designing interactions among agents, tools, models, application services, and data sources.
• Build observability features such as logging, tracing, diagnostics, and monitoring for AI agents.
• Make AI behaviour testable using evaluation approaches, regression tests, and structured test cases.
• Design reliable interfaces with structured schemas and clear contracts between components.
• Integrate AI workflows with enterprise application APIs, databases, and analytical data platforms.
• Operate AI systems beyond prototypes considering latency, cost, failure recovery, token usage, concurrency, model changes, security, and maintainability.
• Build guardrails to control tool usage, data access, prompt injection, unexpected model behaviour, and destructive actions.
• Collaborate closely with software and data engineers to maintain the AI layer as a product component.
• Contribute to architecture and product decisions through design reviews, sprint planning, and feedback cycles.
• Use modern AI development tools such as Cursor and Claude Code or equivalents.

EU residency and permission to work in the EU are required.