Postdoctoral Researcher in Data-Efficient Visual Learning

Teaser Text

The Leiden Institute of Advanced Computer Science (LIACS) at Leiden University is seeking a highly motivated Postdoctoral Researcher to join the Machine Learning cluster. The position is part of a research project investigating how visual foundation models can efficiently acquire new expertise in specialised domains where labelled data is scarce. Recent foundation models have transformed computer vision, yet their ability to acquire new expertise remains limited when training data is scarce or specialised. This project aims to develop the next generation of adaptive visual learning systems by enabling foundation models to efficiently specialise through the exploitation of structure within representations and across tasks. The resulting methods will help bridge the gap between general-purpose foundation models and expert-level performance in specialised domains. The successful candidate will work closely with Dr. Hazel Doughty ([email protected]) and will have the opportunity to help shape an ambitious research programme at the intersection of computer vision, machine learning, and foundation models.

Research Focus
The project investigates how visual foundation models can efficiently adapt beyond their original training experience and acquire expert-level capabilities in specialized domains with limited supervision. Potential research directions include, but are not limited to:
● Adaptation and specialisation of visual foundation models;
● Fine-grained visual understanding and representation learning;
● Self-supervised and data-efficient learning;
● Transfer learning and knowledge reuse across tasks;
● Structured and compositional learning;
● Vision-language and multimodal models;
● Learning with limited supervision;
● Generalisation beyond pretraining distributions The precise research agenda will be developed together with the successful candidate and may evolve throughout the project. 

What you will do

● Conduct original and high-quality research in machine learning and computer vision; 
● Develop novel algorithms for adapting and specialising visual foundation models;
● Publish research findings at leading machine learning and computer vision venues such as CVPR, ICCV, ECCV, NeurIPS, and ICLR;
● Present research at international conferences and workshops;
● Contribute to the supervision of MSc and PhD students;
● Assist with limited teaching and service activities within the department;
● Contribute to the development of an internationally visible research programme on fine-grained and data-efficient visual understanding.

Where you will work

What you bring

We are looking for candidates with:

● A PhD in Computer Science, Artificial Intelligence, Machine Learning, Computer Vision, or a closely related field;
● A strong publication record in leading machine learning and/or computer vision venues;
● Excellent programming skills, particularly in Python and PyTorch;
● Excellent written and oral communication skills in English;
● The ability to conduct independent research while working effectively in a collaborative environment Experience with vision-language models, large-scale representation learning, or foundation model adaptation is  particularly desirable.

What we offer

• An employment contract for a duration of initially one year. After a positive evaluation of the progress of the research, personal capabilities and compatibility, the appointment will be extended by two more years.This contract falls under the CLA of Dutch Universities;
• A salary between € 3.546,= and € 5.538,= gross per month, based on a full-time appointment (38 hours, payscale 10);
• A holiday allowance (8%), an end-of-year bonus (8,3%), and an attractive pension scheme at ABP;
• Full reimbursement of public transport commuting costs for home-to-work travel;
• Flexible working hours: as a standard, you are entitled to a minimum of 29 leave days on the basis of a full-time working week of 38 hours; you can also save for extra leave, for example by working 40 hours a week, and in this way accrue an extra 96 leave hours, or exchange 96 leave hours for a 36-hour week.
• Lots of options when it comes to secondary employment conditions; we can, for example, discuss options for a sabbatical or paid parental leave. Within our terms of employment individual choices model, you can exchange leave days and/or salary for benefits such as an advantageous sports subscription and bicycle scheme;
• If your work allows it, hybrid working is possible within the Netherlands;
• A home-working allowance (day and internet allowance) and attention for good workplaces. The University will also provide you with a laptop.

What we value

Promoting an inclusive community is central to Leiden University’s values and vision. We strive to be an open community in which all students and staff members feel valued and respected, and are able to fully develop themselves. We consider an inclusive academic environment, where students and staff can share their experiences and viewpoints, to be essential for the development of talent.

Want to apply or find out more?