Postdoc In Open-Source Foundation Models

Snapshot

Artificial Intelligence is reshaping our world. Now you get to shape AI. This large and ambitious project aims to develop a groundbreaking family of advanced, multilingual, open source foundation models based on solid research on model architecture, data quality, scaling, generalizability, finetuning, and safety. It combines the unique expertise of leading AI companies and top academic labs to create models with stronger generalization and finetuning capabilities, as well as enhanced transparency and safety. This will be an inclusive, community-driven project designed to and foster a new wave of innovation and scientific advancement.

The team

The Automated Machine Learning team at TU Eindhoven focusses on cutting-edge research to advance the capabilities of machine learning models, while also democratizing AI and leveraging it to help humanity. We are a team of scientists and engineers who aim to deeply understand, explain, and build AI systems that learn continually and automatically assemble themselves to learn faster and better. In addition to producing highly-cited research published at top academic venues, we build models and systems that are widely used by people every day. This work is part of a large and talented team with world-class labs and experts across Europe, supported by well-known companies with advanced knowledge of LLM development.

The role

We are seeking a proficient and talented Research Scientist who is passionate about learning to learn. You will play a pivotal role in a team exploring new model architectures, novel efficient finetuning and adaptation techniques, refining data to enhance learning and generalizability, creating benchmarks, and ensuring safety while contributing to state-of-the-art, human-centric LLM development with real-world impact.

Key responsibilities (depending on your expertise):

As part of a team, you will focus on a subset of the following tasks, aligned with your strengths and interests:

• Research and develop innovative foundation model architectures and scaling laws.
• Explore efficient, cutting-edge finetuning and adaptation techniques based on transfer learning, meta-learning, and continual learning.
• Design and optimize data pipelines for pretraining models to enhance generalization.
• Create new benchmarks to assess safety and generalization abilities across diverse applications.
• Develop tools for continuous evaluation, benchmarking, and training monitoring.
• Collaborate to tackle challenging technical problems and publish research in top venues.

What we're looking for

A self-driven researcher with strong and proven expertise in machine learning. Experience with foundation models, LLMs, pretraining, finetuning, transfer learning, meta-learning, and/or continual learning is a plus. You should have strong technical and programming skills and a drive to create new things. We appreciate a collaborative mindset and eagerness to work with a consortium of leading researchers, companies and stakeholders.

Impact

This is your chance to make a real-world impact by advancing open-source foundation models while collaborating with leading AI researchers and innovative companies. Gain invaluable new skills as you help democratize AI, develop cutting-edge models, and create more efficient fine-tuning techniques that empower responsible innovation and drive scientific progress.

Requirements:

• A PhD in machine learning research, or a comparable domain.
• Ability to conduct high quality academic research, reflected in demonstrable outputs.
• Excellent programming skills.
• A team player who enjoys coaching PhD students and working in a dynamic, interdisciplinary team.
• A proven ability to manage complex projects to completion on schedule.
• Excellent (written and verbal) proficiency in English, good communication and leadership skills.
• A true passion for AI!

Salary Benefits:

A meaningful job in a dynamic and ambitious university, in an interdisciplinary setting and within an international network. You will work on a beautiful, green campus within walking distance of the central train station. In addition, we offer you:

• Full-time employment for 2 years.
• Salary in accordance with the Collective Labour Agreement for Dutch Universities, scale 10 (min. €4,020 max. €5,278).
• A year-end bonus of 8.3% and annual vacation pay of 8%.
• High-quality training programs on general skills, didactics and topics related to research and valorization.
• An excellent technical infrastructure, on-campus children's day care and sports facilities.
• Partially paid parental leave and an allowance for commuting, working from home and internet costs.
• A TU/e Postdoc Association that helps you to build a stronger and broader academic and personal network, and offers tailored support, training and workshops.
• A Staff Immigration Team is available for international candidates, as are a tax compensation scheme (the 30% facility) and a compensation for moving expenses.

38 hours per week

De Rondom 70