Predictive system
Define a prediction target, establish a simple baseline, construct a leakage-safe pipeline, compare models, analyze failure, and explain deployment limits.
- Data contract
- Baseline
- Model comparison
- Error analysis
- Reproducibility
Student Projects
Work the student can explain, defend, and continue.
A Pristone project is not judged by polish alone. The student should own the question, architecture, implementation, testing, evidence, limitations, and next decision.
Free to ask · replies within 2 hours
The Pristone Method
One capability loop across every track.
- Diagnose
- Prioritize
- Build
- Test transfer
- Measure
- Re-prioritize
This page explains the project standard and representative project architectures. It does not present a representative brief as completed student work. Individual artifacts are published only with appropriate written permission and privacy review.
Written and operationally reviewed by Edward Mabonga, founder of Pristone Academy.
Representative briefs
Three kinds of serious project work
These are illustrative structures used to explain the pathway. They are not testimonials or claims about a particular student.
Software product
Start from a user problem, design the system boundary and data model, build an end-to-end version, test it, and document important tradeoffs.
- Requirements
- Architecture
- Implementation
- Testing
- Iteration
Technical investigation
Form a researchable question, select a method, design an experiment, evaluate the result, report limitations, and identify the next investigation.
- Question
- Method
- Experiment
- Evidence
- Limitations
Capability signals
What the finished artifact should make visible
Ownership
The student can explain why each major decision was made and identify where outside help or AI was used.
Architecture
The work has a coherent structure suited to its scope rather than a collection of copied fragments.
Testing
Important behaviors, assumptions, and failure cases are checked rather than trusted because the happy path works.
Evidence
Claims about performance or improvement are connected to a baseline, metric, comparison, or documented observation.
Limitations
The student can state what the project does not establish and where it is likely to fail.
Continuation
The student can identify and begin the next meaningful improvement without waiting for the mentor to prescribe every step.
Privacy and publication
A public portfolio is optional, consented, and reviewed
Written permission
Student work, screenshots, names, quotations, and distinctive project details are not published without the required written consent.
Minor protection
For minors, parent or guardian permission is required in addition to an age-appropriate discussion with the student.
Data minimization
Schools, usernames, exact dates, private repositories, personal datasets, and unnecessary background details are removed.
Ownership retained
The student retains ownership of the student's work. Publication permission does not transfer that ownership to Pristone.
Continue through the Pristone system
Build something that demonstrates real ownership.
A project diagnostic reviews readiness, scope, constraints, and the smallest version that can produce meaningful evidence.
Founder-led by Edward Mabonga (MIT EECS) · no guaranteed scores, admissions, publications, promotions, or jobs