SkillTern

the problem

Computer science doesn’t turn anyone away at the door. It does something quieter: it lets the price of a laptop, the zip code of a school, and the vocabulary of the internet decide who ever walks up to the door at all. The students on the wrong side of that filter don’t experience it as rejection — they experience it as a subject that was obviously for someone else.

That was the part worth breaking. Not the difficulty of the material — kids handle difficulty fine — but the unspoken assumption about who the material was for. A field was quietly deciding who gets in, and the deciding happened years before anyone wrote a line of code.

what I built

My job as CTO was everything that had to exist before a volunteer could stand in front of a student: the curriculum, the digital infrastructure, and the machinery that let strangers teach it well.

• curriculum

  I led curriculum design: computer-science fundamentals and AI literacy, sequenced for students meeting both for the first time. Not a bootcamp syllabus with the prices removed — a path that assumed nothing except that the student showed up.

• infrastructure

  I built SkillTern’s website and a digital certification system, so a student who finished a course walked away holding something they could show. A certificate is a small thing until it’s the only proof you have that you belong in the room.

• partnerships

  We established partnerships with 4+ nonprofits. That is how you actually reach students the field forgot — they are not searching for your landing page, so you go through the organizations already standing next to them.

• volunteers

  I directed volunteer onboarding and training, and standardized the tech stack so every classroom ran the same way. Teaching volunteers to teach turned out to be its own curriculum — and the one that decided whether we could scale at all.

what actually fails

The naïve model of this problem is that the kids lack aptitude and need material. That model is wrong, and building on it is how well-funded programmes produce a lot of certificates and very few programmers.

What actually drops people is friction before the first success. Not the concepts — the twenty minutes of environment setup, the install that fails on the one shared machine, the error message that assumes you already know what a path is. A student who cannot get anything to run has learned something, and what they have learned is that this is not for them. Every minute between a student sitting down and a student seeing their own code do something is the highest-leverage minute in the curriculum, and it is the one nobody designs.

The second thing that drops people is a curriculum built for the teacher's sense of order rather than the student's sense of progress. Ordering by the structure of the subject — types, then control flow, then functions — is elegant and puts the payoff at the end. Ordering by "what makes something visibly happen soonest" is uglier as a syllabus and keeps far more people in the room. I chose the ugly one.

And the third is the volunteer, not the student. Volunteer teachers are capable and inconsistent — they arrive at different times, know different amounts, and leave. A programme where quality depends on which volunteer you got is a programme that cannot grow past the founders' own hours. So the real product was never the lessons; it was writing everything down until a session no longer required a specific person to run it. That is also the only reason it survived me leaving, which was the point from the first week.

the move

You don’t teach 350 kids to code so you can own them. The certification system, the volunteer training, the standardized stack — the common thread is that none of it needed me in the room. That was the design goal, not a side effect. A program that collapses when its founder gets busy was never a program; it was a person with a website.

My time as CTO ended in August 2023. The honest version of this page is that I can’t tell you where each of those 350 students is now — I built a door, not a tracking system, and I think that’s the right trade. What I can tell you is the shape of the move, because it’s the same one everywhere on this site: find the people a field forgot, build the most ambitious thing you can at the gap, then hand them the field and step off.