San Jose, California · Cohorts of five

Do the work before anyone asks you to.

Frontier Scholars is a selective program where high school and university students run original AI research and ship products people can actually use. Students come for work that matters. Families stay because that work is visible, supervised, and theirs to keep.

Request program details See the three levels

Academic research + industry practice Remote or hybrid Application required Three entry levels

What you leave with

Not a certificate. Work with your name on it.

Every level ends in artifacts a student can hand to an admissions reader, a professor, or a hiring manager and let them examine directly. For families, this is what the tuition produces: things that exist, that you can look at, and that outlast the program.

A working product

A deployed AI application or agent, with a public repository anyone can read.

A written paper

A research manuscript or technical report, prepared to the standard of a real venue.

A public presentation

You present your work and defend your decisions to an audience that asks hard questions.

A mentor's assessment

A written evaluation of what you built, how you worked, and where you grew.

The three levels

One program. Three depths of commitment.

Enter at the level that matches where you are. Every level combines research training with hands-on building, and each one is a complete experience as well as a credible route into the next.

Level I

Frontier Studio

Ten weeks of weekend sessions. You pick a problem you actually care about, learn to build with the tools professional teams use right now, and ship a working AI app or agent before the cohort ends.

You leave with: a deployed product, a public repository, and a written case study.

Format 10 weeks
One 2-hour session per week
Weekends, remote or hybrid
Independent build time between

Level II

Frontier Intensive

Two weeks at full pace over the summer. Sixty hours in the room, plus work between days. You go from a blank repository to a launched product and present it at Demo Day, the way a founding team would.

You leave with: a launched product, user feedback, and a recorded Demo Day pitch.

Format 2 weeks
6 hours a day, 10 days
60 contact hours
Summer, remote or hybrid

Level III · Most selective

Frontier Fellowship

Three to six months on one serious question, defined with your mentor. You review the literature, design a method, run the work, and write it up for a real audience. Fellows work toward a public research output: a preprint, a workshop submission, or a peer-reviewed venue where the work is ready for one.

A small number of fellows whose contribution meets the standards of the research team continue as co-authors on collaborative work with faculty researchers. That is earned through contribution, never sold and never promised in advance.

You leave with: a research manuscript, a public output where the work merits it, and a research talk.

Format 3 to 6 months
Weekly mentor sessions
Individually scoped
Remote or hybrid
Entry by portfolio review

Who you work with

People who do this work for a living.

You are not handed a curriculum by a teaching assistant. You are matched to a mentor whose research or engineering work overlaps with your question, and you meet with that person every week.

Our mentors and advisors have conducted research, taught, or trained at institutions including:

Stanford University
Harvard University
UNC Chapel Hill
Santa Clara University
New York University
UC San Diego

The bench grows with each cohort as projects are scoped. Mentors are matched to your project, not assigned in bulk.

Program lead

Hao (Marcus) Xue, PhD is the founder of AI ECD Inc., an applied AI company in San Jose. He was previously a postdoctoral researcher at Stanford University and a data science manager in technology. His academic work has been cited more than 1,700 times.

AI ECD also collaborates with nonprofit organizations working in child and youth development, on selected research projects that fellows may contribute to.

Institution names describe the academic and professional backgrounds of individual mentors and advisors. They do not indicate sponsorship, endorsement, review, or any affiliation between those institutions and AI ECD Inc.

Who this is for

Students who are already restless.

Two kinds of students do well here. Below that, what this looks like from a parent's side of the table.

High school students

You have run out of room in your coursework and you want to make something real before you apply to university. You are willing to be told your first idea is not good enough yet.

University students

You are building toward a graduate application or a technical role and you need research and startup experience you can point to, not a line on a transcript.

For parents

What you are paying for, stated plainly.

Most programs at this price ask you to trust a brochure. Here is what the money actually buys, how you will know it is working, and what we will not claim.

Time with a practitioner, not a syllabus

Tuition buys weekly one-to-one sessions with a mentor who does this work professionally, a cohort capped at five, and a project scoped to your child specifically. There is no recorded curriculum and no teaching assistant standing in.

Progress you can see for yourself

The work is visible as it happens: a repository, a manuscript, a presentation. At the end, your child receives a written assessment from their mentor covering what they built, how they worked, and where they grew.

You are part of the decision

Before anyone commits, we talk with you about which level fits, what the weekly commitment really is, and whether this is the right year for it. We confirm the schedule with your family before a place is held.

What we will not tell you

That this improves admissions odds. That publication is assured. That every student will produce something remarkable. We would rather lose the enrollment than sell you an outcome we do not control.

How admission works

We read every inquiry. We accept very few.

Cohorts are capped at five students because mentorship at this depth does not scale. The steps below happen in order.

Request program details

A short form about the student, their goals, and what they want to build or investigate. Three minutes.

A conversation

Twenty minutes with our team, students and parents together, to work out which level fits, what the commitment really looks like, and whether we are the right match.

Portfolio or statement review

We look at what you have already made or written, then propose a mentor and a project direction.

An offer of a place

If the fit is right, we offer a seat in the next cohort and confirm the schedule with your family.

Straight answers

The questions families ask first.

Frontier Scholars sits deliberately across two worlds: the research training of an academic lab and the delivery pressure of a working technology team. The first questions below cover what that produces. The later ones cover the practical and legal detail.

What will my child actually learn?

Two things at once, which is the point of the program. On the academic side: how to read a literature, frame a question that has not already been answered, design a method, and defend the reasoning behind it. On the industry side: how to build and ship with the tools professional teams use right now, work in a real repository, respond to feedback, and get something in front of users.

Most programs teach one of these. Students who can do both are rare, and that combination is what we select and train for.

How is this different from a course or a summer camp?

There is no fixed syllabus and no recorded lectures. Your child works on one problem they chose, with a mentor who does this professionally, and meets with that person every week for the length of the program.

That means the learning is not simulated. Decisions have consequences, work gets critiqued honestly, first ideas are often sent back, and progress is measured by what actually exists at the end rather than by attendance or a completion certificate.

How does this strengthen a university or graduate application?

By giving a student something specific to point to. Admissions readers see a very large number of applicants who list activities. Far fewer arrive with a deployed product, a public repository, a written manuscript, and a mentor who can speak in detail about how they work.

It also gives a student something real to write about. The strongest essays tend to come from students who have struggled with a problem long enough to have an actual view about it, and that is difficult to manufacture.

We want to be exact here: this is about the quality of what a student can show, not about improving anyone's odds. See the admissions question further down.

Where does this lead after the program?

The same body of work supports two different paths, which is why the program is built across both.

Students heading toward a master's or doctoral program leave with research output and a mentor who can write about their work with specifics. Students heading toward a technical role leave with shipped software, engineering practice, and the experience of building something under real constraints, which is what interviewers actually probe for.

Students at university level often want both open. Keeping both open is the design.

Will this get my child into a top university?

No program can promise that, and we will not pretend otherwise. What we can do is help a student produce work that demonstrates initiative, technical ability, and sustained effort on a hard problem.

University and graduate admissions outcomes are never a condition of enrollment and never a basis for refund.

Will my work be published?

Publication is never guaranteed and never purchased. Fellows work toward a public research output, and where the work is genuinely ready we support submission to a preprint server, a workshop, or a peer-reviewed venue.

We do not use pay-to-publish journals, and we do not list students as authors on work they did not substantively contribute to.

Is this an internship?

No. Frontier Scholars is an educational fellowship. You work on program projects, and the work you produce belongs to you.

If a project later moves toward production use at AI ECD, that is a separate agreement, discussed openly with you and your family.

How small are the cohorts, really?

Five students. A cohort runs with a minimum of three. If a cohort does not reach that minimum, you can move to the next one or receive a full refund of your deposit.

Where do students join from?

Anywhere. Sessions are scheduled around the time zones of the students in the cohort. Hybrid students meet in person in Silicon Valley, California, USA.

What does it cost?

Frontier Scholars is tuition-based, and tuition depends on the level and length. We go through the full figures with parents during the first conversation, before anyone commits to anything and before any payment is discussed.

A limited number of need-based places are available each cohort.

Next cohort

Tell us what you want to build.

Students and parents both start here. Three minutes now, a conversation next. We read every inquiry ourselves.

Request program details

Or write to programs@aiecd.com