%20(1).png){kind=logo}
At Iris.ai, we’re building an agentic AI platform that scales expert-level domain knowledge across entire organizations.
For more than a decade, we’ve worked at the intersection of scientific research, industrial data, and applied AI, helping researchers, engineers, and business teams reason over complex technical knowledge.
Our products - Neuralith, Axion, and RSpace - span the full GenAI lifecycle:
Data ingestion across text, tables, figures, and technical formats
Advanced RAG and indexing pipelines
Agentic orchestration and reasoning
Rigorous LLM evaluation and governance
What makes us different: we care deeply about accuracy, evaluation, and responsibility. We don’t optimize for demos and proof-of-concepts we optimize for systems that experts trust and use.
We’re looking for a Senior/Lead LLM Researcher to lead Iris.ai’s research around LLM evaluation, model interpretability, and mechanistic interpretation of LLMs.
You’ll define and drive research into how we evaluate, understand, and measure the reliability of LLMs - from answer quality and groundedness to uncertainty, confidence, and model behaviour.
This is a research and leadership role with a strong applied focus. You’ll shape research directions, lead experimentation, and work closely with our engineering and product teams to turn research into products used within the Iris.ai platform.
You’ll also participate in shaping projects for EU and national research funding, leading and co-authoring grant proposals such as Horizon Europe and EIC.
You’ll work on a focused set of high‑impact research directions that sit at the core of modern applied NLP\ML\LLM and agentic systems. You will be making LLM-based systems measurable, interpretable, and trustworthy. The core aspects include:
LLM evaluation — model grounding (in context, following instructions, following domain knowledge), faithfulness, and task adherence.
Mechanistic interpretation - analyzing model behaviour during inference. All our metrics are focused on real-time analysis so they require model analysis during inference, not another LLM call.
Model interpretability — understanding and explaining model behaviour and outputs.
Uncertainty & confidence — identifying when outputs can be trusted and when they cannot.
Evaluation frameworks — developing scalable methods for comparing models, prompts, agents, and whole agentic systems
Agentic reasoning & control — understanding when models should reason, stop reasoning, or act, including inference‑time steering
Translation & multilingual NLP — evaluation and system design for modern LLM‑based translation, including low‑resource languages
Your goal will be turning rigorous research into capabilities that real users can trust and use.
Lead Iris.ai’s research direction around LLM evaluation and interpretability
Design novel evaluation methods, metrics, and experimental frameworks
Run rigorous experiments, benchmarking, and ablation studies
Translate research into prototypes and production capabilities
Provide scientific guidance to researchers and engineers
Collaborate closely with product and engineering teams
Publish research and engage with the AI research community
Lead and co-author EU and national research grant proposals
Write and publish research articles
Supervise interns and master thesis students
Languages: Python (strong OOP practices)
ML: PyTorch, Transformers, TensorFlow
LLMs: Hugging Face, OpenAI, custom and fine‑tuned models
Systems: RAG pipelines, Multi-agent frameworks, Evaluation tools
Infra: AWS, Docker, HPC
Practices: Git, CI/CD, reproducible research workflows
PhD in ML, NLP, Computer Science, or a related field
Strong, hands‑on experience with R&D grants and proposal writing (e.g. Horizon Europe, EIC, national or international research funding)
5+ years of industry or applied research experience
Strong background in NLP (transformers, semantic search, RAG)
Hands‑on experience with LLMs and their evaluation
Solid software engineering skills and experience with Python
Publications in ML/NLP conferences or journals
Able to work within European time zones
If you want to do meaningful NLP work, help secure funding for frontier AI research, and grow in a culture built on trust, rigor, and fairness — let’s talk.
We’re not your typical tech company. We believe in:
Real transparency — information is shared, context is open, and questions are welcome.
Fairness, designed in policies, opportunities, and growth are aligned across countries and teams.
Ownership and empoweredness to make decisions without micromanagement.
Metrics that guide us — but they never replace human thinking or responsibility
Pay
Compensation that reflects your value. Our salaries are typically 25% above local market averages, ensuring competitive, fair pay across regions and roles. And we review it annually.
Equity
We believe salary helps you get by. Stock options build wealth. At Iris.ai all colleagues receive ownership in the company, part of our ESOP pool (3%). Because when we grow, you grow — that's what shared success really means.
(Just imagine: Someone once bought a Tesla option for $1 — it's worth $400 today.)
We’ve built our benefits to reflect how we work: with trust, fairness, and room to grow.
30 days paid vacation
5 additional days paid vacation for Learning and Development
Private health insurance (premium coverage) and bi-annual health checks
Free MultiSport card for your physical well-being
Remote-first & flexible hours — work where you're at your best
Personal annual learning budget for conferences, courses, or certifications
Personal equipment budget to choose the gear that suits your style
Charity and volunteer activities
Seasonal working camps (summer & winter) and team retreats
Ongoing growth through weekly tech deep dives, mentorship, pair coding, and knowledge-sharing
If you care about building high-quality, ethical AI — guided by data and human judgment — you’ll feel at home at Iris.ai.
👉 Apply now or reach out with questions. We’re transparent by default.