ETS - AI Research Engineer

Job description

There's a saying at Itineris: Together goes a long way. And we live by that. Whether your job is translating our customers' needs into the best software, selling, coding, or keeping our business running smoothly, we're united by the same mission: growing our company and delivering innovative software that empowers energy and water utilities to engage better with their customers. 

UMAXis the brand name of our software solution for water and energy utilities. It automates their business processes and enables them to better engage with their customers. UMAX is a cloud-based solution, built on the powerful Microsoft Dynamics 365 platform.

As an AI Research Engineer, you will… 
You will join as an AI Research Engineer to our growing AI Practice team. You will own the quality and performance of our AI agent harness — the orchestration layer that powers UMAX's agentic capabilities. Your primary focus is running structured experiments to understand what works, iterating on prompt strategies, tool configurations, and orchestration patterns, and translating those findings into measurable improvements in agent output quality.
This is not a pure research role; your experiments need to ship as improvements. But it is also not a pure engineering role, your instinct for rigorous evaluation and systematic experimentation is what makes you effective.

Key Responsibilities

• Agent Harness Optimization: Own the continuous improvement of our agent harness: prompt engineering, system prompt architecture, tool/function definitions, context management strategies, and orchestration flow design. Form hypotheses, run experiments, measure outcomes, iterate.
• Evaluation Pipeline Development: Design and build evaluation frameworks for agent output quality: scoring rubrics, automated test suites, human-in-the-loop review workflows, and regression tracking. Define what "better" means before you try to get there.
• Experimentation & Analysis: Run structured A/B experiments across prompt strategies, model configurations, retrieval approaches, and agentic patterns. Document findings rigorously. Distinguish signal from noise and translate results into actionable recommendations.
• LLM & Framework Research: Stay close to developments in the LLM and agentic AI space: new models, new orchestration frameworks, new prompting techniques. Evaluate their relevance to UMAX's architecture and bring in what matters.
• Collaboration with AI Engineers: Work closely with AI Engineers to operationalize improvements. You identify and validate what works in experimentation; they build it into production. The handoff between your role and theirs should be seamless.

Required Skills & Qualifications

• Experimentation & Evaluation
  1. Experience designing and running structured experiments on LLM or agentic systems
  2. Ability to define evaluation criteria, build scoring rubrics, and reason about output quality systematically
  3. Comfort working with qualitative and quantitative signals. Not everything that matters is easily measured
  4. Familiarity with eval frameworks and approaches (LLM-as-judge, human evaluation, automated regression testing)
• AI & LLM Knowledge
  1. Deep practical understanding of LLM behavior: prompting, context windows, tool/function calling, retrieval-augmented generation, multi-step reasoning
  2. Hands-on experience with agentic orchestration frameworks (LangChain, Semantic Kernel, AutoGen, or similar)
  3. Understanding of how agent harness design choices affect output quality
  4. Experience with agentic development workflows using tools like GitHub Copilot, Claude Code, or similar
• Software Engineering
  1. Strong proficiency in Python
  2. Ability to write clean, maintainable experiment and evaluation code.
  3. Enough engineering depth to instrument pipelines, interpret logs, and work effectively with production systems
  4. Experience with version control and reproducibility practices for experiments
• Analytical Thinking
  1. Rigorous and curious: you form clear hypotheses before running experiments
  2. Able to communicate findings clearly to both technical and non-technical stakeholders
  3. Comfortable with ambiguity; many of the most important questions in this role don't have obvious answers yet
• Education & Experience
  1. 3-5 years of demonstrable experience as Data Scientist, AI Engineer.
  2. Degree in Computer Science, AI/ML, or a related discipline.

Preferred Qualifications

• Familiarity with Copilot Studio or similar low-code agent orchestration platforms is a plus.
• Exposure to the Microsoft Dynamics 365 / Power Platform ecosystem
• Experience contributing to or evaluating open-source LLM tooling
• Publications, conference talks, or public writing on applied AI topics are a plus

What we offer
 

• A fast-growing and international environment with offices in Ghent (Belgium), Atlanta (USA), London (UK), and Hyderabad (India).
• Working with future-ready water and energy companies and building valuable utility expertise.
• Internal mobility, developing your career through various educational programs and growth paths.
• An entrepreneurial mindset where you have a direct impact on our growth journey.
• An open environment where we understand the importance of taking a break to recharge and achieve the best results. We make it a priority to celebrate milestones, both big and small.
• Onboarding, learning and buddy programs to get ready for a smooth start in your new job.
• A flexible and inclusive working environment – where people can just be who they are, and where we offer flexible options for hybrid working so you can work in a way that suits you best.