Data Scientist Junior H/F
Core
Investigate whether large language models (LLMs) develop biases when trained on American text corpora versus French or European ones, specifically regarding the interpretation of commercial bank communications.
Role type
Junior research intern (NLP/LLM bias analysis)
Builds
Research findings on model performance and bias attribution
Domain
Financial services + Artificial Intelligence (NLP/LLM)
Deliverable
research
Required skills
Deep Learning, Natural Language Processing (NLP), LLM training and evaluation, Python or R programming, statistical foundations
Preferred skills
Economics background, ability to communicate complex technical concepts
Technologies
LLMs, Python, R
Responsibilities
Compare LLM classification capabilities on bank communications using human-validated reference scales, analyze model performance gaps and systematic biases based on training corpus origin, document methodological precautions for future economic and financial analysis
Seniority
Junior, intern