interning at an AI tech company in Hong Kong

On day 2 at a Hong Kong AI startup, the intern already shipped a personal dashboard and a news aggregation pipeline — and called the speed "baffling."

## The Setup Martin interviews Chris, a second-year data science student, on his second day at a small Hong Kong AI company. The team's tight structure is named explicitly: "mainly just me, you, Michael" plus behind-the-scenes staff. Chris is signed on for two months and will get "an entire AI training camp" documented on camera.

## University vs. Real-World AI Chris's blunt take from two days on the job: "in these two days actually applying what I've learned there and learning new stuff has actually equipped me much better for the real world than the university itself." He also flags that universities "never teach you to use" tools like Minimax — "because it's considered cheating." His call to action: universities "have to very much integrate AI and raise the ceiling."

## What He's Actually Building In 48 hours Chris produced a personal dashboard to organize his workflow and started a news aggregation dashboard for AI updates. He routes it through Minimax and stated it's surprising to build "whole pipelines of data" that fast. He attributes the speed to how "much the floor has raised as to what is expected from AI."

## Model Preferences and Prompting Chris leans toward Claude over GPT, and has also tested Kimi and DeepSeek. His rule: "Gemini is used for image creation — you wouldn't ask it to build a whole data pipeline." Half the blame for weak model output, he says, sits with the user: "if you're not engineering your prompt correctly."

## Culture, Not Code The repeated lesson is collaboration. Chris expected isolation and got the opposite — a "laidback but simultaneously more active" environment where "you wear different hats." The team's framing: internal meetings are "sharing XP," because "no one can truly be an expert" in a "self evolving" field.

## Outlook Chris says his AI view "has not so much changed — it has expanded," and admitted the shift comes with fear: AI work is "slightly terrifying" because of "how easily replaceable we also are."