The $12 Americano: An OpenAI Engineer on What Makes Builders Valuable in the AI Era

Korean creator HeeChan Kim has started a long-form series called 시대의 흐름, roughly "The Flow of the Times", about reading big shifts and choosing what to do next. Episode one is a 37-minute conversation with Jay Jung, an engineering lead in an innovation lab inside OpenAI's go-to-market organisation (희야기 | HeeChan, 2026).

The title, description and chapter list are in Korean. The translations in this post are ours. The interview itself is in English, and the points below are paraphrased from its audio. The video is embedded below, and the timestamps jump to the matching chapter.

Who is talking

Jay's path runs through:

• Amazon Web Services, in its compute division;

• Microsoft, building a computer-vision product;

• Facebook, working on APIs for fundraising;

• several startups, including a Y Combinator company;

• OpenAI, where the team builds agent-based products (watch from 2:21).

At OpenAI the team's job is to show how OpenAI uses OpenAI. Customers ask what their "dream" would look like if they used the APIs fully, and the team answers with working tools for sellers and account teams: automated document review, drafted emails, and memory of what was said in meetings (watch from 4:00).

Jay also shares rough pay figures for earlier roles, rising with each move, and declines to discuss OpenAI's. We leave the numbers in the video. They are not the point.

What sets market value

Asked whether value comes from technical skill, scarcity or choosing the right market, Jay answers "all of the above", then picks curiosity as the trait that matters most right now (watch from 6:31). Not curiosity as a personality label. Curiosity about the open questions: how to use what models can do, and how to get that to many people. Valuable people go "the next derivative": they ask how to get there and ship something. Jay thinks this now applies to every job, not just engineering (watch from 7:30).

On why careers diverge, Jay describes two valid strategies (watch from 8:25):

• Go deep and become the expert.

• Go broad, which is what Jay did: get bored once good at something, move on, and end up with cloud, AI and computer-vision experience together.

It is a choice on an axis, and both ends are legitimate.

The memorable image is about markets. An Americano costs about two dollars at a grocery store and twelve at a stadium, and it is the same coffee. Find the place that values your skills at twelve dollars, and keep improving the coffee: stay curious and understand what happens under the hood (watch from 10:02; 10:30).

Reading the wave, and the wall

How did Jay spot AI early? Context. The more you know about a field, the more opportunity you see in it. People who understood the technology knew it would change the world before it reached the public (watch from 11:16).

Compared with the virtual-reality and metaverse hype, AI is simply more versatile. A headset bounds what VR can do. AI spreads into legal work, healthcare and robotics (watch from 12:42).

Jay anticipates the obvious objection: easy to say from inside a frontier lab. Ninety per cent of that career was spent outside one. The advice is to learn as much as you can "outside the wall". With enough knowledge, holes appear that let you in (watch from 14:00).

Inside OpenAI

Jay's definition of a smart person changed after joining (watch from 15:09):

• Curiosity at another scale. Jay thought of themselves as curious. Colleagues turned out to be a hundred times more so.

• End-to-end understanding. Even product people understand how a request travels through the whole system, end to end.

• Learning by default. The people who thrive treat the company's knowledge as something to learn from constantly.

What Jay underestimated was how deeply AI would be woven into ordinary work, from planning and building to communication (watch from 17:42). Jay's rule now is to try automating the parts you dislike first. Jay hands those chores to a coding agent and is surprised when it fails (watch from 18:30).

Robotics and the skills that decay

Asked for something more concrete than curiosity, Jay names robotics. With the time again, Jay would study it more. As a personal bet, robotics addresses a far larger set of problems than VR could: cleaning, dishes, laundry, the work that frees time for people (watch from 19:09; 21:30).

Jay also sees vertical products in law and healthcare, built by people who know both the domain and the model (watch from 20:30).

The sharpest point is about skills. Models improve so fast that a skill you build can be automated by the next release. The durable move is to learn how models work, because that shows you their gaps: what they are not doing yet, and where a company can be born (watch from 22:31; 23:30).

Advice for people starting out

The advice for someone starting out:

• Build things, good or bad. Asked how a student with fifty hours a week should spend them, Jay's answer is to build products, and to ship some of them to real users. That teaches hosting, development and users in a way reading papers cannot. The best engineers Jay works with are usually founders or builders (watch from 24:36; 25:30).

• Find people who grow with you. A partner or close friend who builds alongside you makes learning faster and less lonely (watch from 27:00).

• Don't choose the field for the money. Jay over-indexed on compensation early and regrets it. Great engineers usually end up well paid, and the reverse is rarer, so trust the process (watch from 28:07).

The closing chapters are personal:

• Motivation. Jay wants the person who goes to sleep to know something the person who woke up did not (watch from 30:55).

• The hardest stretch. It was imposter syndrome. Jay worked to turn that distress into the useful kind of stress: there is so much I don't know, how exciting (watch from 33:21).

• One question for everyone. What impact do you want to have? Take a breath and ask it, because life moves fast (watch from 35:33).

Our take for builders

Most of this is classic career advice, said from an unusually good vantage point. Two points are specific to now:

• Skills built around a model's current weakness have a short shelf life. Understanding of the model itself does not.

• The cheapest education available is shipping something small to real users.

The Americano is a useful check, too. If your work is undervalued, the fix may be the market, not the coffee.

References

희야기 | HeeChan. (2026, October 1). OpenAI 엔지니어가 말하는 AI 시대에 몸값이 폭등중인 사람의 특징 | 시대의 흐름 Ep.1 [Traits of people whose market value is soaring in the AI era, according to an OpenAI engineer | The flow of the times, Ep. 1] [Video]. YouTube. https://www.youtube.com/watch?v=-SffBBo40bA