Bell Labs Internship on Data Privacy in LLM-driven Applications (PhD)

at  Nokia

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.). The most capable LLMs (GPT-4, Gemini) are only accessible through APIs. Moreover, the deployment of open-source alternatives requires specialized hardware and software. Therefore, companies often look to use LLMs as a service. However, this raises concerns regarding data privacy and information leakage. For Nokia, a company renowned as a trusted and reliable partner, data privacy is paramount.
In this internship, you will explore strategies to leverage LLMs in a privacy-preserving manner . You will perform a systematic literature review, set up your own experiments, build prototypes and get the opportunity to present your work within and outside the team. Optimally, your work will result in a paper.
Duration: flexible, to be agreed (typically 3-4 months), starting time is flexible

Location: Antwerp (Belgium)

• You will review scientific literature on data privacy in the context of LLM applications.
• You will further explore the most promising approaches, and potentially propose a new approach/a variation on an existing approach.
• You evaluate your technique(s) in the context of a concrete LLM usecase.
• You write a scientific paper based on your work
• Student enrolled in Ph.D. Computer Science/Engineering .
• Programming skills in Python
• Good system building skills=
• Knowledge about machine learning, data mining, information retrieval or natural language processing is a plus.
• Knowledge about data protection strategies (e.g., encryption techniques) is a plus.
• Familiarity with open-source software and tooling (GitHub, Stack Overflow, Pypi, Docker, etc).
• Language skills: English

Please refer the Job description for details