Mark Calleja
Current Work

What I am building and exploring right now.

Some of the many plates I am currently spinning.

Now

Code Club Projects Team, Raspberry Pi Foundation

I currently work with the Code Club Projects (CCP) team, where we design and build coding experiences for young people that are rigorous enough to stand up in a classroom and playful enough to survive contact with actual children. My role lives somewhere between learning designer, creative instigator, and constructive troublemaker, shaping projects, pathways, and toolkits that make computing feel less like instruction and more like discovery.

A particular obsession of mine is exploring how high-engagement, rich learning experiences, the messy, joyful, "wait, can we try this?" kind, can be translated into structures that work inside formal classroom systems. In other words: how do you bottle the weird without flattening it? How do you design it so that any teacher, not just the caffeinated mavericks, can deliver something that feels alive, rigorous, and genuinely empowering? That tension, between freedom and framework, is where most of my work lives. It is part pedagogy lab, part production studio, and occasionally part controlled explosion, and I would not have it any other way.

View Project Vault Read Words
Current work visual

In Progress

Somekone Social Media Lab

Somekone is an experiential media literacy project that turns the mechanics of social media into something visible, testable and discussable in the classroom. Designed for late KS2 and early KS3, it places pupils inside a safe social media simulation where they explore, react and gradually notice how feeds shift over time in response to interaction. Instead of beginning with warnings or technical explanations, learners experience how clicks become data, how engagement drives amplification, and how professionally presented content can spread regardless of accuracy. Through structured discussion, visual network maps and a hands-on "make your own engagement bait" task, pupils build accurate mental models of how recommender systems work without needing coding knowledge or real platform access. The result is a low-risk, high-impact unit that embeds AI literacy within media literacy, foregrounds human choice and design, and equips young people with the ability to recognise attention patterns wherever they encounter them.

Brainrot Brigade

Brainrot Brigade project visual

Brainrot Brigade is an experimental, narrative-driven TTRPG designed to smuggle computational thinking into play. It is a remix of the much-loved one-page RPG Lasers and Feelings , inheriting its elegant, low-friction mechanics and single-number decision-making, then extending them into short, punchy sci-fi adventures built around puzzle-based encounters. Players earn progress through choices, teamwork, and problem-solving rather than stats alone, with each adventure feeding directly into a Scratch-based, modular character sheet they actively build and upgrade over time, adding dice rollers, inventory systems, avatars, and new mechanics as they unlock digital "loot" through play. The result is a tight loop of story -> unlock -> build -> play, where creative coding and roleplay reinforce each other, and learning happens because the game demands it, not because it is labelled as such.

AI productivity and learning

In my current work, I’m poking at the question of how to get the best out of LLMs — in classrooms and in my own slightly overclocked workflow of productivity and tool-building. I’m not especially interested in becoming a prompt whisperer. I’m interested in posture: how we structure interaction so the human keeps their hands on the wheel, even when the machine sounds alarmingly confident.

The Centaur–Cyborg framing is my current lens. In Centaur mode, the model is a powerful but supervised instrument — fast, useful, and cheerfully fallible. In Cyborg mode, it’s woven more deeply into ideation and critique... but never granted authorship. In both cases, the real objective is interface literacy: learning to steer, interrogate, and sanity-check systems that generate fluent prose without guaranteed grounding.

Lately I’ve been experimenting with what I flippantly call friction-maxxing. Instead of smoothing everything into instant answers, I use Socratic prompting to push the cognitive load back onto the human. The model asks questions, surfaces assumptions, and refuses to be sycophantic. You don’t get polish until you’ve earned it, and you usually get far superior content.

The trick is simple: engineered friction produces better and more human-centric content, whatever the purpose. If we design AI interactions that demand articulation, constraint, and reflection, we preserve critical thinking — and build workflows that augment expertise rather than quietly replacing it.

TL:DR - I’m not trying to make AI more helpful, I’m trying to keep humans at the centre of the relationship with these newfangled tools.