Data Scientist | NLP Researcher | Multilingual AI Enthusiast
Geneva, Switzerland

I'm a postdoctoral researcher specializing in large language models (LLMs), cross-lingual/domain transfer learning, and machine translation. My work focuses on making scientific and technical knowledge accessible across languages and domains, with a strong emphasis on open-source contributions and reproducible research.

• Large Language Models (LLMs)
• Cross-Lingual & Domain Transfer Learning
• Machine Translation (MT)
• Information Retrieval (IR) & Retrieval-Augmented Generation (RAG)
• Data Augmentation for NLP

• Languages: Python, R, LaTeX
• Frameworks/Libraries: Transformers, PyTorch, TensorFlow, HuggingFace, Ray Tune, FAIRSEQ, FastAPI
• Infrastructure: Google Cloud, OpenStack, SLURM (HPC)
• Deployment: Docker, FastAPI
• Languages Spoken: English, French, Spanish, Portuguese, German (B2)

• TransBERT: Leading the development of a state-of-the-art LLM leveraging synthetically translated data. Achieves SOTA performance with open-source code and a forthcoming EMNLP2025 paper.
• Large Scale Corpus Translation: Designed a scalable package for translating massive corpora (e.g., 22M PubMed abstracts EN→FR), to be released with TransBERT.
• RAG TREC: Developing a Retrieval-Augmented Generation pipeline for the upcoming TREC conference.

• TransBERT: A Framework for Synthetic Translation in Domain-Specific Language Modeling (EMNLP2025, submission)
• Deep learning-based risk prediction for clinical trials (Patterns, 2023)
• Ensemble of deep learning language models for COVID-19 literature (Systematic Reviews, 2023)
• Multilingual RECIST classification in radiology (Frontiers in Digital Health, 2023)
• More on Google Scholar | DBLP

• LinkedIn
• Email: julien.knafou@gmail.com.com

"Bridging language barriers through AI and open science."