About CodeArchitect

Mission

CodeArchitect is designed for learning the software development toolset needed to create robust, scalable computer systems. Evaluating the performance of key algorithms and data structures, using the syntax and rules for writing the necessary code, and designing modules to compose to create a coherent, usable system. The four introductory assignments — input and output, conditionals, loops, and inserting into a sorted list — are the entry points to that mindset, with more ambitious multi-file projects linked out to GitHub for learners ready to build at scale.

Pedagogy

The goal is an adaptable, engaging platform that centers the core ideas of computer science steady while flexing around the diversity of the people learning them. Drawing from the research literature in CS education and learning science as well as firsthand experience in teaching neurodiverse learners, the site aims to meet learners where they are, and scaffold them into the next step they can take on their own.

Universal Design for Learning

The deeper commitment is to Universal Design for Learning : designing from day one for multiple means of representation, engagement, and action — rather than building a default experience and bolting on accommodations later. In practice this means the content model treats text, code, examples, and (in time) audio and video as parallel ways into the same idea; that interactive controls use real HTML form elements with proper labels and ARIA so screen readers, keyboard navigation, and OS-level assistive tools work; and that animations honor prefers-reduced-motion. Implementing multimedia learning materials is the next key step in CodeArchitect's development.

Zone of Proximal Development

The Zone of Proximal Development is the space where learners are challenged but not overwhelmed. CodeArchitect supports this for now with a simple self directed design that allows users to work through any of the different lessons on the site at their own pace and in whatever order they choose. Adaptive suggestions are a future enhancement if found beneficial.

CS techniques: PRIMM and Parson’s Puzzles

Two evidence-backed techniques support the design of the content. The first is PRIMM: Predict, Run, Investigate, Modify, Make. Rather than throw learners at a blank editor, lessons begin with code that already works. Students predict what it will do, run it to check, investigate how the pieces fit, modify a part of it, and only then build something of their own. This shape is woven through the lesson materials as runnable, inspectable code snippets — so the path from reading to writing is gradual rather than a cliff.

The second is Parson’s Puzzles : students rearrange shuffled pieces of a correct solution rather than write code from scratch. With this support, learners reach equivalent or better outcomes at lower cognitive load and in less time.

Designing for neurodiverse learners

Cognitive Load Theory is treated as a differentiation lever: panels collapse when not relevant to the current step; visual channels do not compete with each other by default; and lesson progress is per-step, so a student who closes the tab returns exactly where they left off to improve executive functioning support. Inline glossary terms surface CS vocabulary on hover for learners who benefit from definitions in place rather than a context switch.

What is implemented today, and what is on the way

Currently in the site:

• PRIMM-shaped lesson flow with embedded, runnable code snippets in the instructional materials.
• Parson’s Puzzles as a first-class assignment type, with indentation and horizontal snippet reordering.
• A progressive hint model — sequential reveal, tracked usage, never an all-or-nothing solution dump.
• An in-browser Python runner built on Skulpt so learners can write and run code without installing anything.
• Reduced-motion support, accessible drag-and-keyboard interactions, ligature-free code rendering, and inline glossary popovers.
• Per-step progress tracking so learners can leave and return without losing context.

On the roadmap:

• Audio narration of instructional text for learners who process information better through listening, and for accessibility.
• Video walkthroughs with live-coding-style narration.
• Six color themes — Midnight, Light, High-Contrast, Rose, Dyslexia-Friendly, and Rainbow — selectable per learner.
• Adaptive sequencing and spaced retrieval — resurfacing earlier lessons when the data suggests it would help — instead of strict linear order.
• Subgoal-labeled worked examples , one of the highest-effect interventions in the CS education literature.
• Localization and translation of instructional content — the content model is locale-aware from the start, but only English is authored today.
• Accounts and saved progress across devices via Supabase, so learners can pick up on any machine.
• A downloadable Diploma summarizing what the learner built, intended as a portfolio artifact rather than a grade.