Summary

Delft University of Technology (TU Delft) is inviting applications for a fully funded PhD Position in AI Alignment: Value Assessment for Open Models & AI Systems. The research focuses on developing methods to assess, quantify, and embed human, ethical, and legal values into AI alignment pipelines for open-weight models and foundation AI systems. The project aims to improve trustworthy AI deployment, particularly in high-risk domains such as law enforcement, in collaboration with the Netherlands Police. (TU Delft Careers)

PhD Position AI Alignment: Value Assessment for Open Models & AI Systems – TU Delft, Netherlands

Subject Area

• Artificial Intelligence (AI)
• AI Alignment and Responsible AI
• Machine Learning
• Human-AI Interaction
• Explainable AI (XAI)
• AI Ethics and Governance
• Foundation Models and Large Language Models
• Algorithmic Fairness and Trustworthy AI

Location

📍 Delft, Netherlands
Faculty of Electrical Engineering, Mathematics & Computer Science, TU Delft. The candidate will also collaborate with the Netherlands Police strategy and innovation division. (TU Delft Careers)

Eligibility / Qualification

Applicants should have:

• A Master’s degree (MSc) in:
  • Computer Science
  • Data Science
  • Artificial Intelligence
  • AI Ethics or another relevant discipline
• Strong background in:
  • Machine learning/artificial intelligence
  • Research methods
  • AI systems evaluation
• Experience through relevant projects, thesis work, or research activities.
• Interest in:
  • AI alignment
  • Human-AI interaction
  • Explainable AI
  • Responsible deployment of AI systems
• Excellent written and spoken English communication skills.
• Knowledge of the Dutch language is required due to collaboration with the Netherlands Police and involvement with Dutch-language stakeholders. (TU Delft Careers)

Scholarship / PhD Description

The selected PhD researcher will work on creating practical frameworks for evaluating whether AI systems align with human, organizational, and societal values.

The research will focus on two major areas:

1. Formalizing AI Value Taxonomies and Alignment Metrics

• Define ethical, legal, and operational requirements for open-weight foundation AI models.
• Develop measurable criteria for concepts such as:
  • Fairness
  • Transparency
  • Explainability
  • Trustworthiness
  • Reliability
• Translate human values into engineering metrics for AI systems.

2. Developing AI Evaluation and Red-Teaming Frameworks

• Design automated and human-in-the-loop evaluation methods.
• Develop AI model auditing and stress-testing approaches.
• Explore:
  • Synthetic data generation
  • Red-teaming techniques
  • Reinforcement Learning from Human Feedback (RLHF)
  • Direct Preference Optimization (DPO)
  • Constitutional AI approaches

The PhD candidate will contribute to safer AI adoption in public-sector applications and help advance practical AI alignment methodologies. (TU Delft Careers)

Funding and Benefits

• Fully funded PhD employment position.
• Salary according to the Dutch Universities Collective Labour Agreement:
  • Approx. €3,059/month in the first year
  • Increasing up to €3,881/month in the fourth year
• Additional benefits:
  • 8% holiday allowance
  • 8.3% end-of-year bonus
  • TU Delft Graduate School membership
  • Research training and professional development opportunities
  • Flexible work arrangements
    (TU Delft Careers)

How to Apply

Interested candidates must apply online through the TU Delft career portal.

Required documents:

1. Updated CV
2. Motivation letter

Applications should be addressed to Dr. Ujwal Gadiraju.

Applications submitted via email or post will not be considered. (TU Delft Careers)

Last Date to Apply

📅 10 August 2026 (TU Delft Careers)

Official Vacancy: TU Delft Careers – PhD Position AI Alignment: Value Assessment for Open Models & AI Systems TU Delft Careers

Link