Large Language Models (LLMs) present an interesting opportunity for Nokia to increase its efficiency and improve the usability of our products (by automating routine tasks, by augmenting human experts in specialized areas, etc.). However, even the most capable LLMs are unreliable: slightly different prompts can lead to very different responses, answers are not guaranteed to be factual (Google's AI overview recently suggested using glue to get cheese to stick to pizza), in multi-user applications LLMs frequently leak sensitive information, etc.
Our team studies algorithms and systems that can facilitate the development of robust and reliable LLM applications. For instance, metrics that measure the confidence that LLMs have in their responses or data provenance tracing algorithms that give insight on what training sources have influenced a prediction (e.g., the ‘pizza glue incident’ was likely caused by sourcing from a sarcastic Reddit post).
Duration: flexible, to be agreed (typically 3-4 months), starting time is flexible
Location: Antwerp (Belgium)

In this Ph D internship, you will do research to help increase the trustworthiness of LLMs for industry applications. Potential research questions could be:

• How can we better measure the uncertainty/confidence LLMs have in their predictions? #callibration #uncertaintyestimation
• How can we better guard against LLMs leaking sensitive information? #dataprivacy
• What algorithms can be used for linking LLM predictions to the training data that influenced these predictions? #dataprovenance

Tasks

• Perform a literature review
• Design & implement algorithms / a system that help improve the reliability of Gen AI applications (see the potential research questions above).
• Evaluate the effectiveness of your approach on representative problems.
• Consolidate your research in a scientific publication.

• Student enrolled in Ph.D. Computer Science/Engineering in Machine Learning
• Strong programming skills in Python
• Language skills: English
• Experience in AI w.r.t. safety, uncertainty quantification, robustness, interpretability, or privacy is a big plus.
• A strong publication record is a big plus.

Come create the technology that helps the world act together

Nokia is committed to innovation and technology leadership across mobile, fixed and cloud networks. Your career here will have a positive impact on people’s lives and will help us build the capabilities needed for a more productive, sustainable, and inclusive world.
We challenge ourselves to create an inclusive way of working where we are open to new ideas, empowered to take risks and fearless to bring our authentic selves to work

What we offer

Nokia offers continuous learning opportunities, well-being programs to support you mentally and physically, opportunities to join and get supported by employee resource groups, mentoring programs and highly diverse teams with an inclusive culture where people thrive and are empowered.

Nokia is committed to inclusion and is an equal opportunity employer

Nokia has received the following recognitions for its commitment to inclusion & equality:

• One of the World’s Most Ethical Companies by Ethisphere
• Gender-Equality Index by Bloomberg
Workplace Pride Global Benchmark

At Nokia, we act inclusively and respect the uniqueness of people. Nokia’s employment decisions are made regardless of race, color, national or ethnic origin, religion, gender, sexual orientation, gender identity or expression, age, marital status, disability, protected veteran status or other characteristics protected by law.
We are committed to a culture of inclusion built upon our core value of respect.