English · 中文
Executive Search — AI for Science & Deep Tech · Energy & Commodities · Financial Data
Shanghai, China · 10 years in executive search, 8 years in-house HR before that.
Most recruiting markets get described from the outside. I work inside the Chinese one, and I write down what I actually see.
I recruit in China for AI for science and LLM teams. If you build models, agents, or the systems they run on, this page is for you.
- Roles I actually work on: ML scientists in materials and molecules · AI4S algorithm and agent engineering · LLM pre-training and post-training · multimodal · agent infrastructure · CUDA and compilers · training and inference systems.
- Where: Shanghai, Beijing, Hangzhou, Shenzhen, and remote-friendly teams.
- I was on the other side. 8 years in-house HR at Sinochem before search. I know how the hiring decision actually gets made, not just how the JD gets written.
- Nothing moves without your permission. No CV leaves my desk, and no client learns your name, without your explicit yes.
- If it is not a fit, you hear it from me. Not silence.
Considering a move? Email me one line about what you build — not a CV. I will tell you what the market pays for it before you spend any time.
I run retained and contingency search for companies hiring in China and across Asia. Two tracks:
1. AI for Science & Deep Tech — the current focus
| Role family | Typical profile |
|---|---|
| ML scientists, materials & molecules | PhD in chemistry, materials or physics; JAX / PyTorch; MLIP, equivariant GNN, DFT |
| AI4S algorithm & agent engineering | LLM agents, agentic workflows, tool use, evaluation, production deployment |
| Self-driving lab / lab automation | instrument integration, orchestration, digital twins, wet-lab fluency |
| AI product & applied science leadership | toB AI products, research-to-product translation, P&L |
| Quant & data science (adjacent) | commodities trading analytics, market data, pricing models |
2. Energy, commodities & financial data — the base
Trading, research, origination, sales and trade execution across oil & gas, power, metals and agriculture; plus market data, pricing, index and analytics businesses.
Client names stay confidential, so here are the categories instead:
| Sector | Counterparties |
|---|---|
| Energy & commodities | international oil & gas majors, commodity trading houses, metals & mining producers, power and renewables developers, agricultural traders |
| Financial data & analytics | price reporting agencies, index and benchmark providers, market-data and analytics vendors, research houses |
| AI for Science & deep tech | materials and drug discovery startups, lab-automation and self-driving-lab builders, AI infrastructure and agent-product companies |
Completed engagements can be discussed privately. Live searches, clients under NDA and candidates are never disclosed.
I automate my own desk. Everything here is a working protocol I use daily, stripped of client and candidate data.
| Repo | What it solves |
|---|---|
| big-plan | An agent skill that refuses to let you drop a lead. Multi-turn projects die by silent skipping, not by wrong answers — this enforces a checklist the agent is not allowed to abandon. |
| keyword-pairing | Sourcing method: turn a job description into Anchor × Pairing Term search pairs, so a search returns 20 right people instead of 2,000 wrong ones. |
Three things I hold myself to, and they show up in the tools above:
- The agent gets equal standing. If a lead is unresolved, it must say so and refuse to move on until I have done my side. A tool that silently drops your work is worse than no tool.
- Every claim carries its evidence grade. Verified / high-confidence / inferred / not obtained. I would rather say "I could not read it" than guess.
- No ghosting. Candidates and clients get a real answer, including when it is no.
- LinkedIn — the fastest way to reach me
- Email: michael.qian@riverstoneassociates.cn
If you are a candidate: one line about what you build, and the roles you would actually consider. No CV needed to start.
If you are a client: retained search, market mapping and talent-pool projects in AI for Science, deep tech, and energy/commodities.
michael.qian@riverstoneassociates.cn
Skills here are MIT licensed and built for Claude Code, Codex, or any agent that reads SKILL.md.