PUBLIC PREVIEWAvailable to explore
PathTutor
Enter a math problem and solution as text or an image, then request an AI review of the work. Check any analysis against the original problem and steps.
AI PRODUCT STUDIO · LEARNING & ASSESSMENT
We turn learning and assessment problems into
real product workflows.
RyongWorks is an early-stage studio building AI products and prototypes around real user workflows. PathTutor has a public preview; AI Interview remains in design and development.
AI analysis can be wrong. Review PathTutor results against the original work. AI Interview is not yet released.
CURRENT WORK
Project status
PathTutor
Public previewA learning tool for reviewing math problems and solution steps entered as text or an image.
Open previewAI Interview
In developmentA product direction shaped around job-specific evidence, clear criteria, and human review.
See design principlesWe distinguish available demos, prototypes, and plans. A public preview is not a claim of validated performance or outcomes.
PRODUCTS
What you can try and what we are building
We label product status clearly and separate public interfaces from features that are still being designed.
IN DEVELOPMENTDesign and development
AI Interview
We are shaping a product for interview and skills assessment around role- and seniority-specific evidence, comparable work samples, transparent criteria, and human review.
View design principles →HOW WE WORK
Build small. Make limits clear.
Start with a workflow
Define whose work the product should support before choosing technology.
Show evidence and limits
Separate previews, prototypes, and plans. Do not present unverified outcomes as facts.
Keep people accountable
In assessment, prioritize transparent criteria and human review.
OPEN SOURCE
Open source projects
Public source repositories and demos, with project scope and limitations stated clearly.
PUBLIC REPOSITORY
ContextFidelity
A local-first, open-source chat workspace that selects context for each question and preserves the exact context sent to each model call.
It does not measure model attention or establish how much a context item influenced an answer.
Source repositoryPUBLIC REPOSITORY · PROTOTYPE
BuildProof
An evidence-based AI technical interview and project review prototype. With candidate consent, a minimized assessment manifest and its salted hash can be recorded in a Solana Devnet Memo transaction.
This records the submitted manifest, not an independent certification of skill. The original resume and full interview transcript are not written on-chain.
ABOUT RYONGWORKS
Explore the studio and its project notes.
We distinguish products, prototypes, and future plans.
Have a product or collaboration question?
Tell us what you are working on at contact@eunryong.com.