Web

SaaS

Aicade 2D Game Builder

Full Stack Engineer — game builder frontend, FastAPI AI services, and API integration

AI-assisted 2D game builder: React/Vite/Next.js editor frontend plus FastAPI services for generation, asset recommendations, and procedural content into a Phaser runtime.

Overview

Aicade lets creators turn a game idea into a playable Phaser 2D experience. The product surface is a high-interaction editor (Vite/React and Next.js product pages) with Redux for editor state and TanStack Query for server data. Generation, asset recommendation, and procedural content run through FastAPI backends that wrap model providers and return structured payloads the editor can apply to the scene graph.

Problem

Prototyping a 2D game usually means hand-building assets, levels, and wiring before anything is playable. Creators needed an editor that could propose assets and structure while staying responsive under heavy client state.

Solution

Owned frontend architecture for the builder and product UI, and FastAPI services that expose generation and recommendation endpoints. The editor consumes those APIs, applies results into Phaser scenes, and keeps Redux + Query caches coherent during long-running generation.

Frontend

Interactive 2D builder UI: Vite React editor, Next.js product surfaces, Redux for scene/editor state, TanStack Query for FastAPI-backed resources.

React

Vite

Next.js

TypeScript

Redux

TanStack Query

Tailwind CSS

Phaser

Level/editor canvas with Phaser for playable preview and interactive level design
Redux slices for selection, tool mode, entity trees, and in-progress generation jobs
TanStack Query for listing assets, polling generation status, and invalidating after FastAPI completions
Next.js surfaces for marketing/product entry; Vite app for the heavyweight editor tooling
Optimistic UI and loading states so AI latency does not freeze the canvas
Typed API client mapping FastAPI JSON schemas into editor-friendly models

Backend

FastAPI (Python) services for AI game generation, asset recommendations, and procedural content—REST endpoints consumed by the React editor.

FastAPI

Python

Pydantic

Uvicorn

LLM / AI provider APIs

REST + JSON schemas

FastAPI routers for generate-game / recommend-assets / procedural-level style operations with Pydantic request/response models
Async handlers calling model providers, normalizing outputs into structured game/asset DTOs for Phaser
Status and result endpoints so the frontend can poll long-running generation via TanStack Query
Validation and error envelopes that the editor can surface without crashing the scene
Separation of product Next.js BFF-style calls from the Vite editor’s direct FastAPI client where latency matters
Environment-based provider keys and timeouts tuned for interactive creator workflows

Modules

Game editor

Canvas, tools, entity inspector, and Phaser playable preview

AI generation

Prompt-to-structure flows backed by FastAPI generation routes

Asset recommendations

Suggested sprites/audio/tiles returned as typed asset payloads

Procedural content

Level/layout suggestions applied into the editor scene graph

API & integration highlights

FastAPI POST generate — structured game/level payloads from creator prompts
FastAPI GET/POST assets — recommendation and retrieval for editor libraries
FastAPI job/status — poll long-running generation for TanStack Query
Editor applies API DTOs into Redux + Phaser without full page reloads

Challenges

Keeping editor UI responsive while FastAPI generation jobs are in flight
Stable contracts between Pydantic schemas and TypeScript client types
Coherent Redux scene state when AI overwrites large subgraphs
Balancing Vite editor tooling with Next.js product surfaces

Impact

Faster idea → playable 2D prototype path for creators
AI assets and procedural content inside the same editor loop
Clear FE/BE split: React editor ↔ FastAPI AI services ↔ Phaser runtime

Architecture

Next.js product + Vite React editor

Redux + TanStack Query

FastAPI AI services (Pydantic)

LLM / asset providers

Phaser 2D runtime

Tech