One person amplified into a team — we hold ourselves to that too.
We build one product: an AI COO for cross-border commerce. Three business lines run on the same capability and the engineering headcount is six — that headcount only works because every person's output has been amplified by AI. It is not a bonus point. It is a condition of joining.
We look at three things
How fast you learn matters more than what you already know
At this stage the business keeps moving. We hire you for one thing and a year later it is probably something else — so what we screen for is not what you know but how fast you pick things up. Knowledge can be filled in; the inability to learn cannot. In the interview we will hand you something you have never done and watch how you break it down, what you ask, and how you get around it when you are stuck. No trivia questions.
You are willing to sit across from the customer
Our customers are cross-border commerce companies, and engineers get embedded in their offices to work out requirements face to face with management and operations leads. Requirements are not written in a document waiting to be collected; they are asked for while sitting in someone else's office. We will ask you to walk through something you built, then give you a real merchant request to repeat back — to see whether what comes back is the words or the understanding.
You already use AI to amplify yourself
Fluency with tools like Claude Code is a requirement. We will look at how you actually use it: how you organise context, how you get it through long tasks, what you no longer write yourself. Not whether you say you have used it.
Open roles
All roles below are full-time and based in Shanghai; the research internships are at the Hong Kong University of Science and Technology.
- Engineer · Storefront automation2
- Turn cross-platform operating actions into automation that holds: fix what breaks, build what real merchants ask for, finish the publishing pipeline.
- Engineer · Enterprise capability2 (one front-end leaning)
- Take it from working well for one person to working well for a company: accounts and permissions, org-level sharing of knowledge and skill libraries, dashboards.
- Product manager1
- Own the whole path from requirements through to design, alone. All three business lines route their requirements through you and land in one product. Producing the designs yourself is best; if not, you need to be able to drag them out of an outside designer.
- Finance (and operations admin)1
- Run the books across multiple entities on your own, dealing with bookkeepers, auditors and banks. Multi-company multi-storefront accounts, the basis and calculation for bonus sharing, and the contracts and equipment for everyone joining.
- Operations lead1
- An interface role, not a large operations team. Work with partner cross-border teams and get day-to-day execution onto the ground.
- Product intern2
- Master's candidate who can write code. Take well-specified issues and run alongside the product.
- Research intern (PhD)3
- PhD candidates working on agentic systems, co-supervised with a lab at HKUST. One direction each: evaluation benchmarks, self-evolving agents, and model post-training. The goal is papers, not near-term product.
Pay and incentives
Twelve months' salary in full, no performance discount
Base is base. It is not scaled down by a performance multiplier. This is the one thing we have fixed about compensation.
Bonuses follow your line's operating results
No common pot. The operating result of your business line determines that line's bonus pool; horizontal roles are tied to the three lines combined. It is added on top of base, not carved out of it.
How to apply
Send your CV to the address below with the role in the subject line. Include one piece of real work you did with AI — what the job was, how you organised it, what it saved. That paragraph tells us more than the CV does.
hello@theglobaloptima.com