Senior AI Engineer
Summary
Contract B2B role (12 months, fully remote in Poland, €400–425/day) building an agentic AI assistant inside a retail promotions optimization platform. Day to day: designing agent workflows and tools, sandboxed code execution, MCP servers, and LLM evaluation/observability — primarily in Python with Pydantic AI/LangChain, Azure OpenAI and DuckDB/SQL.
Senior AI Engineer (Agentic Systems)
Contract: B2B, 12 months with view to extend
Rate: €400–425 per day
Location: Fully remote (Poland)
Build the AI assistant behind a retail promotions platform.
Our client is a globally recognised, award-winning firm, consistently ranked among the world’s best places to work. They are building an AI assistant inside their retail promotions optimization software. It helps users explore data, answer analytical questions, build tables and charts, and generate evidence-based promotion recommendations.
What you’ll do
- Design and build agent workflows that explore datasets, run analyses and answer questions, owning agent composition, tools, skills, message history and conversation storage
- Connect agents to existing application services through typed, governed tools, so answers match what users see in dashboards
- Build sandboxed, resource-limited execution of model-generated code, with no access to credentials or source databases
- Produce robust Plotly charts and tables that follow application styling and number-formatting rules
- Make prompts, skills, tools and data definitions configurable and versioned per client, without forking the core code
- Build MCP servers and clients so agents across the product portfolio can share data and collaborate
- Own evaluation and hardening: scenario-based evals, reconciliation checks, tracing and observability (Phoenix, OpenTelemetry), quality/latency/cost baselines, guardrails and least-privilege data access
- Partner with product, design and client-facing teams, write clear architecture and decision documents, and mentor teammates on agentic engineering
What you’ll bring
- 6+ years in software engineering, data analytics or a related field, including 2+ years in agentic engineering
- Strong Python, with production-quality, well-tested services
- Hands-on experience building agents with Pydantic AI or a comparable framework (LangGraph, LangChain, OpenAI Agents SDK)
- Solid grasp of core agentic patterns: tool calling, multi-step orchestration, context management, guardrails and human-in-the-loop
- Experience with MCP and multi-agent or cross-application agent communication
- Experience building agents that perform data analysis: exploring data, generating and executing code, interpreting results
- Experience designing configurable, extensible systems (per-client variation, plugin or skill patterns, versioned and testable configuration)
- Understanding of data security in agent workflows, including least-privilege access and failure handling
- Practical LLM evaluation (offline evals, LLM-as-judge plus deterministic checks, regression testing) and production monitoring of quality, latency, cost and failure rates
- Experience with LLM observability (Phoenix, OpenTelemetry or similar)
- A track record of owning features end to end and working with product and client-facing stakeholders
- Daily use of AI coding assistants (Claude Code, Codex or similar), with all output validated before production
Strongly preferred
- Pydantic AI (including the Pydantic AI Harness)
- LangChain (the current solution is built on it and will be migrated away from)
- DuckDB, Ibis, pandas or SQL
- Azure OpenAI and Docker
- Dash and Plotly
Nice to have
- Background in retail, CPG, pricing or promotions analytics
- Experience taking AI features from prototype to production in a client-facing product
- React and FastAPI
Skills
Sign in to read and join the discussion.