AI PRODUCT STUDIO · LEARNING & ASSESSMENT

Turning learning and assessment workflows into verifiable AI software.

RyongWorks builds interactive learning tools and evidence-based evaluation software. We currently operate the PathTutor public preview—combining a 3-chapter AI math & ML learning simulator with a multimodal solution review note—while developing AI Interview.

Launch PathTutor Explore products

01 / Workflow Interactive & Stepwise

02 / Evidence Source-Backed Review

03 / Boundary Explicit System Limits

SYSTEM DIRECTORY

Current Products & Releases

2026.10

PT-01

PathTutor

Public Preview

Interactive 3-chapter ML/DL math simulator (/lab) paired with a multimodal math solution analysis note (/tutor).

/lab/linear /lab/logistic /lab/mlp /tutor

Open live app ↗ Architecture & scope →

AI-02

AI Interview

In Development

Skills-assessment workflow structured around role-specific rubrics, comparable work samples, and human reviewer accountability.

Design principles →

PRODUCTS

Products and Interactive Workflows

Each product page documents the live routes you can inspect today alongside explicit operational boundaries.

PATHTUTOR · LEARNING Public Preview

PathTutor

Combines an interactive 3-chapter AI math & machine learning simulator (/lab) synced with 4-stage PyTorch code and a multimodal math solution analysis note (/tutor) powered by Cloudflare Workers and Gemini.

01 Ch.1 Linear Regression · Ch.2 Logistic Classification · Ch.3 MLP Backpropagation
02 4-Stage live code execution flow (PyTorch / TS) & localStorage experiment snapshots
03 Handwritten image & text math solution step review with multi-path comparison

Open public preview ↗ View scope & notice →

AI INTERVIEW · ASSESSMENT In Development

AI Interview

Designed around role- and seniority-specific rubrics, comparable candidate work samples, and transparent evidence summaries that assist—never replace—human interviewers.

01 Role- and seniority-specific structured prompts and work samples
02 Transparent rubric mapping tied directly to candidate transcript evidence
03 Human-in-the-loop review workflow without automated rejection claims

Read design principles →

OPEN SOURCE & PROTOTYPES

Open Source Repositories & Prototypes

View all repositories →

CONTEXTFIDELITY Public Repository

ContextFidelity

A local-first chat workspace using SQLite FTS5 and BM25 keyword retrieval to select relevant context per query and preserve the exact payload sent to each model call.

Node.js · SQLite FTS5 · BM25 GitHub ↗

BUILDPROOF Public Repository · Prototype

BuildProof

An evidence-based technical interview prototype that lets candidates optionally anchor a minimized assessment manifest and salted hash to a Solana Devnet Memo transaction.

Cloudflare Workers · Solana Devnet

GitHub ↗ Live Demo ↗

COLLABORATION & INQUIRIES

Have a product question or collaboration proposal?

Reach out directly by email with the workflow or project you are exploring.

contact@eunryong.com About Studio