Avalon

Lead Game Designer

MAR 2022 - JUN 2026 • CREATE AVALON CORP

Avalon is a community-created MMORPG blending the depth of tabletop RPGs with the scale of a persistent online world. Players explore realms, conquer dungeons, craft equipment, and battle in fast-paced skill-based combat, all while wielding the same creation tools used to build the game itself to design their own levels, stories, and worlds for others.

I joined as Avalon's first designer and was promoted to Lead after the Vertical Slice. I was the vision holder for Creator Mode, the in-game toolset players use to build with modular kits, generate worlds, populate them with NPCs, and author gameplay logic without leaving the game. I led the world design of Elysara, and designed the harvesting and crafting systems underneath it. Avalon reached closed alpha before I left.

The Mandate

Avalon had to be built with the tools we gave players. Not similar tools. The same ones. If a tool didn't meet our needs, it wouldn't meet a player's.

That decision set the pace for four years. Tools had to come online fast and iterate faster, because anything that slowed us down was blocking our own development while it sat in the backlog. Every production problem was a usability test with a deadline attached.

It also meant designing for two audiences through one interface, with no advanced mode to hide complexity in: a professional team building an MMORPG, and a player placing their first wall.

500,000+

individual blocks placed in Elysara by the development team — using the same tools every player gets.

Major Contributions

Vision holder for Creator Mode, and full owner of it after the Vertical Slice

Led world design of Elysara: 6 km², 3 biomes, 11 zones

Designed the magnet snapping system and the block standards behind it

Designed harvesting and crafting, from PCG tooling to recipe economy

Designed Merlin, Avalon's no-code, card-based logic system

First designer hired: co-authored the GDD, mentored designers and environment artists

Systems Design

The snapping system I built, broke, and rebuilt

Object placement was the first problem to solve, and the one everything else depended on. Engines give you free placement or grid snapping. I rejected both. Grid snapping is precise and it feels like software, and we had committed to creation feeling auto-magical. So I designed magnet points instead, borrowing the idea from vertex snapping in 3D modeling: attach points to a mesh, and when two points come near each other, the objects pull together.

It worked. Then it broke at scale. Because magnets were assigned per asset, blocks that should have paired cleanly didn't line up, and walls of different sizes left gaps in a house you couldn't close. Assigning them by hand was hours of work per asset for our own team.

The system wasn't wrong. The unit was. Magnets were per asset when they needed to be per type. After I was promoted to Lead I rebuilt it around a standard: every object type gets fixed dimensions and fixed magnet positions, refined by slope, size, and style. A 45° stair and a 26° stair are different profiles. A corner stair and a straight stair are different styles. But every 4×4 straight wall gets its magnets in exactly the same place, whatever it looks like.

Two things fell out of that which I hadn't fully planned for. An engineer and I built an Editor Utility Widget that reads an asset's tags and assigns its magnets automatically, so anyone on the team could ingest assets. And because magnets no longer belonged to individual meshes, kits became interoperable: a player could build a house from the Village kit and swap every component to the Dungeon kit without rebuilding it. For a multi-genre game, that mattered more than the fix did.

Demonstration of Creator Mode being used to create a village POI

Demo of the Castle Kits modular blocks

Hours → Minutes

Per-asset magnet assignment

1 Day

Full kit ingestion, down from days

Interaction Design

Rebuilding the interface around the game it lives in

The Vertical Slice HUD borrowed from Minecraft: a hotbar, and an overlay screen that covered the thing you were building while you picked what to build it with. Tools lived on key combinations that were getting harder to remember every time we added one. Both were reasonable early calls. Both were wrong, and taking over the feature meant I got to say so.

Avalon's combat doesn't use a hotbar. It uses the number keys for abilities. So I bound Creator Mode's tools to the number keys too. Duplicating a selection is the 3 key, reached the same way you'd reach for an ability. Two halves of the game, one motor memory. Then I moved every menu into a sidebar: kits, NPCs, the Marketplace, each expanding into panels of categories and pieces. That kept the middle of the screen empty, so you can watch the thing you're making while you make it.

Selection needed the same treatment. Drag-select is fine with a mouse in an engine and translates badly to gameplay, worse on a controller. I designed a 3D volume you place in the world and stretch until it contains what you want. Everything inside gets selected. It reads as an object in the world rather than a gesture on a screen.

The same rework carried the Prefab System: group any selection, save it, edit it, share it. By alpha, more than 200 developer-created prefabs had been curated into the game for players to build with.

Creator Mode being used to create a small cathedral

Creator Mode HUD in Closed Alpha

25

Simultaneous builders per world

200+

Developer Prefabs created

System Design

Turning game logic into an object you can pick up

Creator Mode solved building. Logic was the harder problem, because the same rule still applied: if it needed a manual, it wasn't done. Node graphs are the standard answer, and they're the wrong one here. A graph gets more intimidating exactly as it gets more capable, because there's no edge to it. Twenty nodes of wires crossing wires, reading like the code it was supposed to hide.

I designed Merlin around a smaller unit instead: the card. A card holds three menus, Properties, Logic, and Display, and scopes to one behavior at a time. Logic runs on events and actions: assign On Trigger Enter, choose Spawn Monster, tune the conditions underneath it. A card reads like an object because it is one, something a player already knows how to hold and place, not a diagram to interpret.

Cards only render in Creator Mode, so logic never clutters play. When a piece of logic needs a body, it attaches to a building block instead: a chest that spawns loot when you open it, a torch that triggers a trap, a switch that starts a timer. And because a card is an object, it shares like one, the same as a prefab. Bundling logic this way meant a player could chain several cards into a working game loop, spawner to trigger to timer, without ever meeting the word "script."

Merlin Mockup: Monster Spawner with Card Properties exposed

Merlin Mockup: Monster Spawner Card with all tabs in view

World Design

Elysara, and leaving room for the players in it

Elysara is Avalon's first world: 6 km² of high-fantasy island torn off a greater landmass and stranded in the sky, with fractured geography and collapsed edges. I led its design end to end: three biomes, eleven zones with NPC level bands and resource spawns tuned to each, every major and minor POI, the pathways connecting them, and the critical path running west to the capital, south through the archipelago, and back northeast to the Tower at the center. I was also the team's HLOD and PCG subject-matter expert, keeping a world that size performant.

I personally built the starting path: a cliffside, an ancient dungeon, then the reveal of Dawnmere and the world past it. Controls, combat, quests, harvesting, and building all get taught inside that sequence, and none of it announces itself as a tutorial.

The harder problem was restraint. A UGC world has to feel authored enough to show players what's possible, and empty enough that they can claim a piece of it. The obvious move is to mark out build zones. I inverted it: we marked Restricted Zones instead, covering only the developer content we didn't want touched. Everything else was open by default, which left far more of Elysara in players' hands than a permit system would have.

I also built Avalon's first dungeon myself, entirely in Creator Mode: three chambers, connecting halls, and a final arena against a skeleton lord newly returned from death. It set the bar for dungeon content, and it was the most honest test the toolset ever got.

Environmental showcase of Elysara, Avalons High Fantasy themed world

Player overlooking Elysara

6 km²

World Size

11 / 3

Zones, Biomes

System Design

Harvesting, and Crafting

Resources had to feel like they belonged where they grew, across 6 km² of world. Hand-placing them would have taken forever and still looked deliberate. So I built a blueprint on Unreal's PCG that spawns harvestables onto any static mesh. A designer picks a log, picks what grows on it (mushrooms, fiber, bark, clay, iron, moss), then tunes density and spawn rules: underside only, nothing on steep faces. Artists and designers could scatter resources organically without asking engineering for anything.

Under that I set up the harvestable content itself: 24 trees you can chop, with damage thresholds and yields tuned by species and size, 8 mineable rocks, 20 gatherable plants. Then I distributed all of it across biomes and PCG graphs so density read correctly zone to zone.

The crafting economy sat on data tables I authored: where each resource lives, how it's harvested, which workstation holds which recipe, and what every building block costs. Block costs only apply inside a claim during gameplay. In Creator Mode, building is free. Building as play and building as creation are different activities, and charging for both would have punished the thing we most wanted people to do.

BP I created that enabled the spawning of harvestable content on SM assets.

Harvestable Trees in the Meadows biome

24 / 8 / 20

Harvestable Tree, Rocks, and Plants

Leadership

Solving the problem behind the solution

I was the design team's primary line to the C-Suite for four years at a fully remote company. Most of that job was translation: leadership would arrive with a list of wants, sometimes with a solution already attached, and I'd turn it into a plan for how we'd actually get there. I learned not to argue with the solution. I'd find the problem underneath it and solve for that instead, which almost always landed better than either the idea they brought or the objection I would have raised.

I designed in options rather than single answers, so leadership had real places to weigh in early instead of reacting late. The Restricted Zones call in Elysara and the free-building rule in Creator Mode both came out of exactly that process. Alongside it, I mentored designers through 1:1s and reviews, and managed environment artists across Elysara's POIs, Mini Meadows, and Vireon, a 3 km² cyberpunk desert I led through pre-production.