Hi! I’m Chen, a research scientist at the Technology Innovation Institute, working on large language models, natural language processing, and information retrieval. Before that, I completed my Ph.D. in Computer Science at the Weizmann Institute of Science , where I was advised by Prof. Robert Krauthgamer and Prof. Boaz Nadler. During my Ph.D., I interned at Booking.com and K Health, where I got to apply my research to personalized travel and healthcare chatbots.

These days, I’m especially interested in how we can get GenAI systems to provide more reliable, useful, and accessible information.

I’m always happy to talk about research, feel free to reach out!

Updates:

  • July 2025: New preprint on arXiv! We show how language bias in retrieval-augmented generation (RAG) systems can arise from cross-lingual retrieval issues and suggest simple ways to reduce it.

  • May 2025: Our paper on quantifying the distracting effect of irrelevant passages on LLMs has been accepted for oral presentation at ACL 2025. Talk video is available here!

  • July 2024: I joined TII as a Research Scientist! I'm working on the intersection of artifical intelligence and search in the AI/IR research center led by Yoelle Maarek.

  • May 2024: I defended my PhD! You can find my thesis on the research page.

  • January 2024: Our paper on efficient methods for sparse learning in distributed settings has been accepted to AISTATS 2024 in Spain!

  • September 2023: I started an internship at Booking.com as a Machine Learning Scientist, working on personalization within Gen-AI chatbots for travel planning.

  • August 2023: I received the Yoel Meches Memorial Fellowship, also awarded by KKL-JNF.

  • August 2022: I've been awarded the Moshe Meir Horovitz Memorial Fellowship by KKL-JNF.

  • March 2022: I had a great time speaking at WiDS Tel Aviv! In my talk, "How a thousand little failures can turn into success", I discussed how, even when each individual device contributes only limited and noisy information, we can still learn from distributed data. The recording is available here.