LLMs Customization for Automated Vehicles Validation

Project Description:
The development of realistic driving scenarios plays a crucial role in the validation of automated vehicles (AVs). This project explores how large language models (LLMs) can be utilized to enhance scenario design. These scenarios enable testers to assess how AVs perform in various driving conditions, which is vital for ensuring safety and reliability.

Responsibilities:

• Learn how to fine-tune different LLMs for various types of datasets to optimize the generation of realistic driving scenarios.
• Evaluate the effectiveness of fine-tuned LLMs using different algorithms to assess their accuracy and reliability in scenario creation.
• Engage in the entire customization process, from initial concept to final implementation, providing insights and feedback to enhance model performance and scenario accuracy

Your profile:

• A strong interest in the intersection of AI and real-world applications
• Experience or coursework in AI, machine learning (practical experience in LLMs is a plus)
• Upper intermediate programming skill with Python

Our offer:

• Possibility of choosing this topic for your thesis or working as a HiWi
• Opportunity to work part of the project at Royal Holloway, University of London, engaging directly with project partners 
• Challenging tasks in applied research and development in close contact with industry
• Contribute to cutting-edge research projects with close supervision and the potential for scientific publications
• Extensive flexibility (Remote/On-site) for good harmony with your student obligations
• Possibility for long-term HiWi contract up to 80 hours/month

How to Apply:
Interested candidates should submit the following application materials via email to majid.jegarian∂kit.edu with the email subject: [Thesis or HiWi - "Candidate Surname" – LLM-2]

• Resume
• Academic transcript