Qurbit Research Group

Independent research, learned by doing it

Computer Science · Finance · Medicine

Applications read as they arrive

Frontispiece. Harmonograph, drawn continuously

Abstract

We mentor high school and college students through a first piece of independent research: sixteen weeks on a project that already exists, supervised by the person responsible for it, finished in writing. There is no fee, no prior experience is required, and roughly half of accepted students arrive without any.

Apply for a placement Read how it runs

Accepted work

Work from our projects has been accepted at

NeurIPSNeural Information Processing Systems
IEEE IROSIntelligent Robots and Systems
AAPM&R
APHAAmerican Public Health Association
ACSM Annual Meeting 2026American College of Sports Medicine
Annual Southeastern Pediatric Research Conference
International Conference on Emerging Infections and Public Health
Preventive Medicine ReportsJournal

With more submissions currently under review

1What this is

Qurbit is a research group that teaches independent research by having you do some, supervised. We work on problems in computer science, finance, and medicine, and we keep seats on those projects open to students, in high school or college, at no cost.No application fee and no program fee. We do not sell admissions consulting to families afterwards either, which is where a fair number of “free” programs make their money.

The arrangement is unusual enough to be worth explaining. Most research programs aimed at students are courses. You pay a few thousand dollars, get assigned a topic that exists because it was assignable, and write it up toward a certificate that was never in doubt. The result is a paper nobody reads, including the person who supervised it. Whatever that teaches, it is not how to do research on your own.

Our projects were running before any student joined them, and that cuts in both directions. The work is real, with a deadline somebody cares about and a result that might come out wrong. It also means we can only take students when a project has room for one, which is why we turn down people we would have been glad to work with.If we decline you, we will say which of the two reasons it was. Capacity is the more common one.

2What a placement looks like

A cycle runs sixteen weeks. Most of the calendar goes to reading and to analysis, and three weeks are held back for writing.

Figure 1. Phases of a placement cycle. Boundaries move by a week or two depending on the project. The writing block rarely moves, because it is the part students most often underestimate.

The writing surprises people who assume it is a formality once the work is done. It is where most students find out which of their results they cannot yet explain, and it is the part of the process nobody has taught them.

Your mentor scopes the work so it can finish, reads your methods before you spend six weeks on something unsound, marks up your drafts, and tells you when a result is too thin to submit anywhere. They do not write it for you.Worth saying because the incentive to do so is real. A supervised paper that looks impressive is easier to produce than one a student can defend.

3The piece you own

You are not handed a project. You are handed one question inside one, cut down until it can be finished properly in a cycle rather than gestured at in three.

Figure 2. A placement sits two levels down. The project outlives your cycle, which is what makes the deadline real and the scope narrow.

4Areas

You pick one of these when you apply, and the specific project comes later, at matching.The list is short because we only open an area when someone here can supervise it properly. Adding a fourth would mean finding that person first.

Say which topics interest you and why. A student who wants to work on one particular thing is easier to place well than one who is open to anything.

01Computer Science

Algorithms, systems, and computational methods, including work that leans heavily on implementation and measurement.

Algorithms and complexity · Machine learning · Software systems · Quantum computing · Data and simulation

02Finance

Markets, risk, and quantitative methods, with an emphasis on questions that can be tested against real data.

Quantitative modeling · Market microstructure · Risk and portfolio analysis · Financial data science · Behavioral finance

03Medical

Health, biology, and clinical questions, approached through data analysis and careful study design.

Biomedical data analysis · Public health studies · Computational biology · Medical imaging methods · Clinical research design

5Where the work ends up

Finishing a project and publishing one are separate events, and we keep them separate when we talk to students about it.

Figure 3. Outcomes of a completed cycle. The common case is the first one, and it is a real result rather than a consolation.

6Terms

Item	Terms
Fee	No cost to students or families
Time	5–8 hours per week, for one cycle
Length	One project cycle runs about 16 weeks
Format	Remote, with weekly mentor check-ins
Eligibility	In high school or college, anywhere in the world. No prior research needed.
Authorship	Credited by contribution, settled in conversation before anything is submitted.
Publication	Where the work supports it. Most cycles end in a complete internal writeup instead.

Anything not covered here is worth asking about before you apply rather than after.

7Applying

The form asks for two written answers, and those carry most of the weight. Everything before them is administrative.

Figure 4. From application to start. We reply either way, and a decline comes with a reason.

Start an application Ask something first