What is an intern trusted to ship by the end of a summer, and what remains outside that trust? This note draws the line, using a research tool that checks a public volume file as the example, across New York, London, and Hong Kong.
A tool is a finished object
An internship is not a tour and not a shadow of someone else’s password. The useful version ends with an object a reviewer can run: a check, a parser, a notebook that makes a data problem visible, or a test that fails when a log is incomplete. Shipping means that object is in a place other people can find, with a note that says what it does and what it refuses to do. It does not mean the intern’s name is on a live order. Live capital decisions stay with the firm. That sentence is the boundary of the trust, in every office.
Interviews exist partly to see whether a candidate wants that kind of object. A candidate who describes success only as being near the market is preparing for a different summer from the one this note describes. The summer is for building something small and true. New York, London, and Hong Kong can all host that work. The trading day they look onto differs. The meaning of a shipped tool does not.
An intern is trusted to ship a reviewed research tool with a written limit. An intern is not trusted with a live capital decision. That split holds in every office.
A check on a public volume file
A concrete example is a tool that reads a daily volume file for New York Stock Exchange names and flags holidays stored as zeros, tickers that do not map, and days whose totals jump because a condition code was included or dropped. The tool does not decide whether volume is a useful feature. It makes the file’s surprises legible. An intern can own that tool from the empty repository to the reviewed note. A researcher can then accept or reject the file. Acceptance stays with the researcher. The intern has still shipped something real.
The example is deliberately short of an order path. An intern who is invited to wire the tool to a system that can send orders has been invited past the boundary, and the invitation is the error. Interviews should have described the boundary already. If they did not, the intern is entitled to ask. Asking is not a failure of enthusiasm. It is the same instinct the volume tool is supposed to encode: say what the input is, and say what the program must not do.
Offices change the neighbors, not the object. An intern in Hong Kong may finish the day as New York opens. An intern in London may overlap both. An intern in New York may sit inside the U.S. cash session. None of those facts turns a volume check into a marketable order. Mentors in each office are expected to keep the object on the research side of the wall. A summer that produces a careful tool and no blurred credentials is a successful summer.
Trust is specific
Trust, in an internship, is permission to finish a defined tool and to have that tool reviewed as if it mattered, because it does. It is not permission to speak for the firm, to promise a result, or to carry a decision about capital. Interviews that imply otherwise should be corrected by the work itself. The three offices are one standard. A reader comparing New York, London, and Hong Kong should expect the same boundary and different time zones.
What interns are trusted to ship is a research tool or a supporting piece of code with a written limit. The volume-file check is the picture. The interview is where that picture should already have been honest.
This piece does not describe a particular intern’s project, a hiring target, or an offer of employment. It states the boundary for internships in New York, London, and Hong Kong. A check on a public volume file is the example of something that can be shipped without crossing the boundary. Shipping that check is a complete summer outcome, not a prelude to a market instruction.
