Start with curiosity
Discovery gives a new learner a place to explore before choosing a formal learning path.
Open DiscoveryCase study / Software engineering / Independent project
DURA: learning that stays with you
Dustin Snellings’ open-source learning platform: curriculum, offline progress, and an engineering approach grounded in verification.
- My role
- Curriculum and platform author
- Approach
- AI-assisted implementation; explicit verification
- Evidence
- Public product, source, and regression tests
01 / The problem
Learning rarely happens in one sitting.
A learner may lose their connection, switch devices, or return to a lesson later. DURA is an open-source learning platform built around a simple expectation: the work a learner has done should remain useful when they come back.
I built the curriculum and platform, directing AI-assisted implementation of offline progress, spaced repetition, and in-browser code execution. My manufacturing background informs the approach: define what should hold, then inspect whether it does.
02 / Explore the product
Try an idea before choosing a path.
Product walkthrough / Public DURA screens
From “let me try”
to “I understand.”
A short tour of the real product. Screens captured September 12, 2026; open each destination to try it live.
Make the idea tangible
Algorithm Kitchen lets learners arrange instructions and observe what the sequence does. It connects an everyday task to programming concepts.
Try Algorithm Kitchen03 / Decisions and tradeoffs
Make the behavior inspectable.
Progress belongs to a lesson.
Lesson identity includes its phase and module. Two lessons named “01” must not share a completion record.
Tradeoff: a longer identifier carries context that a short display number cannot.
A late result must not replace current work.
Loading and completion can overlap with navigation. The public test suite covers an earlier completion finishing after another lesson has opened.
Tradeoff: asynchronous state needs explicit checks beyond the normal click-through path.
Offline capability adds responsibility.
The public project supports local progress and optional cross-device sync. A local-first experience also needs clear rules for restoring, resetting, and reconciling state.
Tradeoff: continuity for the learner requires more lifecycle handling in the implementation.
04 / Read the evidence
A claim should have somewhere to lead.
Lesson identity and asynchronous progress
The regression suite includes same-number lessons across modules and phases, a reset canceling a pending load, and a completion finishing after navigation. These tests document specific failure cases; their presence alone is not a claim that every production scenario has been validated.
Try the simplified progress demonstration
The portfolio’s progress challenge illustrates keeping the higher progress value when reconnecting. It uses sample state, not a live learner account or DURA’s complete synchronization system.
Open the interactive progress challenge ↗05 / What this demonstrates
Working software, with clear boundaries.
This case study shows product implementation, state-management decisions, and public regression evidence. It does not claim measured learning gains, adoption, or business outcomes. The walkthrough images are dated captures; the linked product and source continue to evolve.
Public evidence reviewed September 21, 2026: DURA’s project description, repository, and the progress files linked above.