@Phoenixyin13: Nanjing University CS LAMDA Group 2025 Graduates Employment: A Microcosm of the High-Purity Involution Era in AI. As the top face of domestic machine learning, the career destinations of Professor Zhou Zhihua's LAMDA Group at Nanjing University for the class of 2025 directly strip away the last arrogance of the current AI job market. First, the visible trend is the academic circle completely closing its doors...

Graduates of Nanjing University’s LAMDA Group (Class of 2025) — a microcosm of the hyper-competitive AI era.

As a flagship face of domestic machine learning, Professor Zhi-Hua Zhou’s team at Nanjing University’s LAMDA Group has seen its 2025 graduates’ destinations directly strip away the last shred of pride from the current AI job market.

First, the visible trend is an all-out closure of academia. Even among LAMDA’s PhDs, only one became a lecturer at Zhejiang Gongshang University (a non-92 institution), while another stayed as a postdoc at Nanjing University — essentially an academic buffer. Many PhDs have abandoned academia entirely, flooding into industry: Huawei (4), Alibaba (2), ByteDance (1), Optiver (1). Nowadays, even top ML PhDs find it astronomically hard to land a faculty position at a good university. The academic slots are completely locked down.

Two companies swallow half the pie. Looking closely at the list, Huawei and Alibaba form the largest cluster. In the past, LAMDA graduates would carefully handpick core teams at big tech firms — like early Tencent AI Lab or Baidu IDL. Now they go in batches to Huawei, Alibaba, ByteDance, and Kuaishou. This shows that even master’s students from top labs, facing today’s employment winter, can only treat stable headcounts at big companies as the safest refuge.

As I said before, for some AI people, the final stop has to be Wall Street or financial districts to grab money. Top-tier foreign quant firm Optiver snagged one PhD; domestic leading quant firms Ubiquant and Mengxi Investment each took one master’s. For the cream-of-the-crop algorithm brains, switching to high-frequency trading and using code to “harvest” the market is the fastest liquidity and highest-return dimensionality reduction move.

Star unicorns can’t absorb many people. Everyone is hyping large models and AGI, yet the whole table shows only one person went to Moonshot AI. High hype, few slots — this indicates that while leading domestic LLM startups have high valuations, their R&D teams are extremely lean, picking candidates to the bone, and simply can’t wholesale talent like traditional internet giants.

We also see that pursuing a PhD in AI after a master’s is a deliberate, calculated decision. It’s different from going straight from a US bachelor’s to PhD. Everyone has their own trade-offs and decisions.

Phoenix Yin (@Phoenixyin13):
Employment of Zhejiang University’s Master of Finance (Class of 2025)

The same bank and the same internet company appear repeatedly, indicating fierce competition and everyone crowding into hot slots.
Graduates of 2025, in the macro environment, still prefer certainty. Internet companies are present but not overwhelmingly dominant.
The gap between headquarters/ministerial positions/top-tier internet vs. local banks/grassroots government tiers/ordinary internet is still stark.