Summary
Jingtao Han is an LLM researcher based in Shanghai with a decade of experience applying deep learning and graph methods to real-world problems across e-commerce, document understanding, and finance. He has moved from research internships at CFM to engineering roles at 4Paradigm and Shopee, where he built multimodal OCR+NER systems and marketplace item-profiling GNNs, and most recently focused on training large language models for math at Project Numina. Trained at Imperial College and École Polytechnique with a background in theoretical physics, he blends strong mathematical foundations with production-minded engineering. Comfortable spanning research and applied engineering, he repeatedly ships models that bridge NLP, vision, and graph modalities—an approach reflected in his concise GitHub persona, “Chic à la jône.”
10 years of coding experience
7 years of employment as a software developer
MSc with Distinction, Advanced Computing, MSc with Distinction, Advanced Computing at Imperial College London
Bachelor of Science (BS), Theoretical Physics, Bachelor of Science (BS), Theoretical Physics at Nanjing University
Suzhou High School
Ingénieur X-2013, Mathematics, Ingénieur X-2013, Mathematics at Ecole polytechnique
English, Chinese, French