Independent machine-learning research for ambitious students
Question23 gives you the structure, tools, and research mentorship to go from curiosity to literature, experiments, code, writing, and submission—while keeping the research yours.
Start with Topic Exploration and mentor guidance before choosing a plan. No research or machine-learning experience required.
Why Question23 exists
“I built the research environment I wish I had.”
Built from the process behind an independent high-school machine-learning project accepted for publication in Frontiers in Earth Science: Cryospheric Sciences.
Frontiers does not endorse or sponsor Question23.
Question23 research desk
Interface preview · no student data shown
Experiment ledger
Experiment title
Status
1. Plan the comparison
OpenQuestion this run answers
Hypothesis or expected comparison
2. Record the method
Next3. Save results and decide
NextThe real ledger ties data, split strategy, metrics, code, results, figures, interpretation, and the next decision to one dated experiment.
One workspace from first question to final paper
Research does not happen in five disconnected apps. Neither should yours. Question23 keeps what you read, decide, test, build, write, and revise inside one continuing research record.
01
Ask the mentor
This is not another blank chat. The mentor works against the continuing project record, teaches what becomes useful, challenges the method, and turns each meeting into a concrete work plan.
02
Run the experiment
The Experiment Ledger keeps the hypothesis, dataset version, methodology, split, metrics, results, figures, interpretation, and next decision together and dated.
03
Build the implementation
Research Lab can turn a saved experiment into a reviewable code proposal, run the approved implementation, and keep its outputs with the project. Missing choices come back as questions instead of being silently invented.
04
Write where the research lives
Write or import a draft, preserve immutable versions, request structured research review, and keep every criticism visible until you decide how to resolve it.
05
Get it ready to submit
Choose a venue when the paper is ready, check its guidelines, prepare the submission, and work through reviewer feedback round by round.
Mentor · continuing project context
Bring something to show
A result, a draft, progress, or a blocker. The mentor responds using the research record that already exists.
Explain what changed, ask why a next step matters, or challenge a suggestion that does not fit the work.
Code workspace · Research Lab
analysis.py
Saved in this project
# Build from your saved experiment
# Review every proposed file before applying
def prepare_data(source):
"""Keep the split boundary explicit."""
return sourceManuscript workspace · versioned writing
manuscript.md
Working draft
# Introduction
Connect the motivation to the decision your result could change.
## Methodology
Explain the full process so another careful researcher could follow it.
Project record · decisions that persist
Recorded
Sources
Recorded
Experiments
Recorded
Versions
Important context is tagged and carried forward so the project does not reset every time the mentor responds.
Living research map
The map shows what the project needs without pretending research moves in a straight line. Reading can change the question. Results can change the method. Writing can reveal another experiment.
Topic exploration · The Question
Literature · Steelman & Novelty
Evidence · Preservation · Method · Implementation
Validation
Write-up · Venue
Review Loop
Founder case study
In high school, I wanted to do serious machine-learning research on avalanche danger. I developed an original methodology around a very limited dataset. There was no recipe: some of the most important work was deciding which physical assumptions could make the problem feasible without making it scientifically meaningless.
My mentor understood research and machine learning, but not snow science. That was valuable. He did not give me the scientific answer. He challenged the methodology, asked for evidence, examined the results, and helped me determine what needed to happen next. I still made the intellectual decisions.
The work was accepted for publication in Frontiers in Earth Science: Cryospheric Sciences. Question23 grew from one realization: much of a great research mentor's transferable value is process expertise, not already knowing your subject.
Your question will look completely different. The process won't.
Founder pain → product infrastructure
I did not know what came next
Living Research Map
My mentor knew research, not snow science
Persistent Mentor
Experiments became hard to distinguish
Experiment Ledger
Decisions were buried across files and calls
Daybook + Project Record
Implementation had to match the method
Code Workspace
Writing exposed holes in the research
Manuscript + revision loops
Acceptance created a new workflow
Submission + reviewer response
Mentorship without substitution
AI can generate an answer. Research requires a process.
The mentor teaches, critiques, explains, asks for evidence, and helps you choose a useful next step. It can disagree with you—and you can push back and ask why. It does not get to quietly make the scientific decisions for you.
You own
The mentor contributes
Research that keeps moving
Independent projects often disappear between bursts of motivation. Question23 creates a flexible weekly rhythm, records the next deliverable, and helps you recover when life—or the research—gets in the way.
Next mentor meeting
A result. A draft. Progress. Or a blocker.
Manuscript → publication process
Question23 reviews the manuscript as research, not only as writing. It looks for methodology gaps, unsupported claims, weak evidence, missing figures or tables, unresolved criticism, and risks that could cause a quick rejection.
01
Draft
02
Review
03
Revise
04
Venue
05
Submit
06
Reviewer feedback
07
Revise again
Submission isn't necessarily the end of the project. Neither is Question23.
Venue-specific checks and reviewer rounds become new, accountable revision work.
College application value
The value is not merely saying “I did research.” It is being able to explain the problem, why it mattered, what failed, what changed, what the evidence showed, and what remains uncertain.
Question23 helps students do work they understand well enough to defend. It does not guarantee admission.
For parents
See whether the project is active, what the next commitment is, when it is due, and why work was postponed—without reading private mentor conversations, drafts, or results.
See the parent experience →Parent progress view
Private research stays privateContinue the project
Your saved Topic Exploration becomes the beginning of the paid project—not a form you have to repeat.
Complete environment
$99 / month
The complete guided research environment.
For machine-learning research
$149 / month
The research environment plus the technical workspace for machine-learning projects.
Questions before you begin
No. Topic Exploration starts with interests and accessible reading. The mentor teaches research and machine-learning concepts when they become useful, while letting a confident student move faster.
No. It helps you explore possibilities, understand existing work, and test whether an idea is researchable. The direction and final question remain yours.
No. The mentor can explain, critique, ask questions, and provide structured feedback. You own the scientific decisions, interpretation, claims, and manuscript language.
No. Optional learning modules explain core concepts, metrics, experiment design, and problems such as train/test leakage. Research Lab can also help implement the experiment you have designed.
You can save an initial interest, explore possible directions, and talk with the mentor before choosing a plan. Your exploration remains saved if you leave the checkout page.
Research is the complete guided research environment. Research Lab adds the technical code workspace, experiment-grounded code generation, code runs with preserved outputs, file and PDF analysis, and higher usage limits for machine-learning projects.
Parents can see practical progress: commitments, due dates, schedule changes, recent work, and whether the project is active. They do not receive the student's private mentor conversation, draft text, or research results.
No. Question23 helps a student conduct, understand, and present serious work; it cannot guarantee a journal decision or an admissions outcome.
Question23 supports venue selection, a venue-specific readiness check, submission preparation, and new revision rounds if reviewers respond. A finished draft is a milestone, not necessarily the end.
Your research project can start with an interest
Explore what interests you, see what people have already tried, and begin shaping a direction with the mentor.
Explore your idea free →