AI Engineer

XTB is a global company from the financial industry, focusing on online trading of financial instruments. It is the largest FinTech in Poland and a leader in Central and Eastern Europe, operating in several countries including Asia and South America.

• 2-5 years of commercial experience in AI/ML or Software Engineering with production deployment.
• Ability to write modern, Pythonic code with solid understanding of AI Agent architectures, including context engineering, tool calling, skills, and agent loops.
• Hands-on expertise in developing Model Context Protocol (MCP) servers or custom tool-calling interfaces.
• Practical experience with LLM orchestration and data frameworks such as LangGraph or LangChain.
• Experience building and maintaining production-grade APIs (REST, gRPC, MCP).
• Solid understanding of database technologies (PostgreSQL), vector databases (pgvector), and advanced search strategies (semantic, hybrid, BM25).
• Familiarity with MLOps/LLMOps concepts, observability tools (MLflow, Langfuse), automated evaluations, and prompt optimizations.
• Strong communication skills and ability to collaborate in an Agile, cross-functional team environment in English.
• Experience monitoring application health using Prometheus and Grafana.

Nice to have:

• Experience in AI Agent evaluation combining classic ML practices and GenAI best practices including LLM-as-a-judge calibration.
• Knowledge of financial systems, trading, or regulated fintech environments.
• Experience securing AI applications through LLM red teaming, adversarial testing, and anti-jailbreak policies.
• Familiarity with cost and latency optimization such as context compression and prompt caching.

• Implement and maintain end-to-end AI solutions, focusing on the robustness and scalability of LLM pipelines and agentic workflows.
• Develop and optimize RAG systems and AI agents using modern frameworks, ensuring high-quality retrieval and response accuracy.
• Build and integrate production-grade APIs (gRPC, MCP) to expose AI capabilities to the global trading platform and internal business units.
• Measure agent performance using LLMOps standards (agent tracing), MLOps practices (automated evaluations), and engineering metrics like latency and error rates.
• Apply evaluation-driven development practices with frameworks like RAG-eval, LLM-as-a-judge techniques, and model benchmarking to optimize system performance.
• Collaborate with Product and Data teams to build and maintain MCP servers and data-serving layers that empower AI agents with real-time enterprise data.
• Maintain high code quality through unit tests, linters, type-checkers, documentation, and code reviews.
• Use agentic coding tools (Claude Code, Codex) to increase engineering efficiency and share learnings.