Agentic AI Engineer

Andersen is hiring an Agentic AI Engineer for a project integrating AI capabilities into a sports performance platform and delivering scalable, reliable insights for coaches and athletes. Our customer is a global technology provider specializing in performance monitoring, data analytics, and operational optimization. It develops software platforms and data-driven solutions that help organizations improve efficiency, gain valuable insights, and make informed decisions. Serving clients across international markets, the company focuses on innovation and integrated technologies that enhance performance and support long-term growth. The project is focused on developing and integrating agentic AI capabilities into a sports performance technology platform. It aims to transform athlete data into actionable, trustworthy insights for coaches and performance specialists through reliable, scalable, and measurable AI-powered solutions.

Responsibilities: - Designing and shipping specialist AI agents that use memory, tools, data and multi-step reasoning. - Building multi-agent orchestration that routes work between agents, manages dependencies and synthesizes conflicting outputs. - Developing mechanisms that assess confidence, uncertainty and consequence before a recommendation reaches a practitioner. - Building human-in-the-loop escalation so the system decides when to answer, when to ask for more information and when to defer to a human. - Turning sport scientist expertise into validated, versioned and testable agent capabilities. - Building evaluation, observability and regression testing to measure and improve agent performance in production. - Working with domain experts to ensure AI outputs are grounded, traceable and actionable. - Developing AI capabilities that match comprehensive athlete data to deliver fast, evidence-based performance insights and recommendations. - Ensuring recommendations include measurably calibrated confidence levels and transparent supporting evidence, allowing practitioners to understand and verify their origins. - Designing AI systems that recognize when available evidence is insufficient and avoid providing unsupported recommendations. - Supporting informed, defensible practitioner decisions while ensuring final decision-making responsibility remains with the human expert.

Requirements: - Professional experience in Applied ML, AI, or Software Engineering for 5+ years. - Strong proficiency in Python and solid software engineering fundamentals, including building, deploying, and operating production systems. - Hands-on experience personally building production agentic AI systems used by real users, including persistent memory/state, tool calling, multi-step reasoning, workflow orchestration, and production operations. Experience limited to chatbots, prompt engineering, or RAG is insufficient. - Hands-on experience building or significantly contributing to production multi-agent systems, including agent routing, orchestration, specialist agent composition, and dependency-aware workflows. - Strong understanding of parallel and sequential agent execution, conflict resolution between agent outputs, and response synthesis. - Practical experience with probabilistic model calibration, including Platt scaling, isotonic regression, Expected Calibration Error (ECE), reliability/calibration curves, and confidence and uncertainty estimation. - Level of English – from Upper-Intermediate and above.

Nice to have: - Experience with production RAG pipelines, retrieval optimization, and reranking. - Experience with foundation model fine-tuning or domain adaptation using LoRA, PEFT, or similar approaches. - Experience with LLM observability, drift detection, evaluation frameworks, and automated regression testing. - Knowledge of human-in-the-loop architectures, confidence thresholds, and escalation mechanisms. - Understanding of causal and counterfactual reasoning. - Experience with Go and AWS services, including ECS, EC2, Lambda, SNS, and SQS. - Experience with GraphQL, REST, or gRPC APIs and databases such as PostgreSQL or MongoDB. - Experience collaborating with sports scientists, clinicians, or other domain experts. - Familiarity with workload, readiness, recovery, biomechanics, or athlete performance data.

Compensation: 3050–5250 EUR per month.