Sun Tianxiang Joins Baidu as Head of Foundation Model R&D
N.R. Finch
Sun Tianxiang, a 27-year-old large-model researcher who built the open-source ChatGPT-class model MOSS, has joined Baidu to head its Basic Model Unit (BMU) and sit on its Model Committee — the latest move in Baidu's push to bring young technical talent into its AI core.
Who is Sun Tianxiang, and why does he matter?
Sun was born in 1997. During his PhD, he led the development of MOSS — one of China's earliest open-source large language models, comparable to ChatGPT.
He coined the MaaS (Model-as-a-Service) concept — treating a large model as a cloud service — and has published over 40 papers at top AI venues including ICML, ICLR, and NeurIPS.
This means → he is not a pure academic. Stints at Amazon AI Lab and the Shanghai AI Laboratory gave him hands-on experience spanning NLP research through large-scale model training and engineering.
What exactly will he do at Baidu?
Baidu now splits its large-model R&D into two parallel units: the Basic Model Unit (BMU) and the Applied Model Unit (AMU).
Sun leads BMU, which builds the foundational model itself. Jia Lei heads AMU, which turns models into products.
In plain terms = BMU builds the engine; AMU builds the car. Sun has been given the most upstream seat in Baidu's model stack.
Why is Baidu making this move now?
Over the past year, Baidu has been reshuffling its AI organisation: it created the Baidu Model Committee (BMC) and recruited multiple cohorts of young technical talent.
Sun also joins the BMC. This means → he will not just run day-to-day R&D for one unit but will shape top-level model-strategy decisions.
This reflects a shift at the foundation-model level — from an internally promoted leadership bench toward bringing in young, externally proven technical leaders.
What is the market still watching?
Before joining Baidu, Sun founded a startup called Rixingji. Han Rui, its first-round investor, described him as "absolutely someone with fire in his belly."
Put simply = he is a builder with entrepreneurial instincts, not a researcher who waits for instructions.
Whether his academic depth and product sense can form real synergy with Baidu's existing foundation-model strategy is the key test observers will use to judge this appointment.
Content is for reference only, not financial advice.