Florian Zumbrunn

I build interfaces where humans direct generative systems.

12+ years shipping creative tools, rendering engines, and now AI-driven workflows.

Why Runway Labs

For over a decade I've been the person between design and engineering. The one who prototypes the idea before anyone else can see it, who builds the internal tools that unblock the team, who translates what a designer envisions into something that actually runs in a browser. I'm comfortable speaking both languages, and I've always enjoyed being the bridge.

At Fable I built the rendering engine and designed the UI for the features it powered. As a freelancer I shipped award-winning interactive experiences that required both creative direction and deep technical execution. As an artist I build my own editors, my own generative pipelines, my own production systems, because the tools I need don't exist yet.

AI has amplified all of this. I now ship in hours what used to take weeks. I build interfaces where humans direct generative models with spatial, visual control, not just text prompts. From a 3D scene editor that lets users compose environments and send them to an AI model, to multi-model pipelines and AI-powered creative workflows I deploy for clients. My toolkit has expanded dramatically, but the core skill is the same: turning a vague creative intent into a working system, fast.

Reading about the Design Engineer role at Runway Labs felt surprisingly familiar. Prototyping new interaction paradigms for world models, working at the frontier between research and product, building tools that don't exist yet. This is the work I care about most, pointed at the most exciting problem in creative technology right now. I'd love to help build it.

Selected Work

Fable — Browser-Based Motion Design Platform

Core engineer on a browser-based animation platform ("Figma for animation"), designing and shipping the WebGL rendering pipeline, shader system, and creative tooling UI.

Fable editor showing timeline, canvas, motion design tools and layer properties
Node-based workflow system with Translate, Text and Transform nodes
Chromatic aberration and tritone effects panel with halftone pattern preview
Layer transitions panel showing Fade, Slide, Scale animation presets
Shape properties panel with Origin, Transform, and Layout controls
Connection panel with Fill, Stroke, and Appearance controls
Role Senior Creative Software Engineer (2019–2023)
Stack WebGL, GLSL shaders, React, TypeScript
  • Designed and shipped the rendering pipeline: shaders, particle engine, post-processing stack
  • Owned UI/UX implementation for new features end-to-end — from prototyping interaction patterns to shipping production components
  • Contributed to product vision, directly shaping how animators interact with creative tools at scale

3D Scene Editor — AI-Directed Environment Generation

A Three.js scene editor that lets users compose 3D scenes and send them to a generative AI model that redesigns the environment while preserving the hero subject.

Project under NDA — workflow overview below.

Workflow diagram: user places elements in 3D editor, scene sent to AI model for processing, environment regenerated while preserving hero subject
Role Lead developer — designed and built the full interface
Stack Three.js, React, generative AI model integration
  • Built for a major advertising agency to accelerate creative production
  • Users arrange hero products and elements in a 3D scene, then the AI model regenerates the surrounding environment — lighting, textures, context — while keeping the hero subject intact
  • A direct example of spatial, intuitive control over generative AI output — beyond text prompting

Hennessy X.O. — Generative Art System

A generative art collaboration with Hennessy X.O., creating algorithmic artworks produced across multiple mediums through an autonomous creative pipeline.

Hennessy-branded robotic arm drawing on a cognac bottle with generative artwork in background
Framed generative artwork with painted Hennessy X.O. bottle displayed below
Three unique generative artworks paired with their corresponding painted bottles, robotic arm at right
Role Artist & creative technologist
Stack WebGL, custom generative algorithms, robotic drawing systems
  • Created a complete generative system: algorithms generate compositions, an autonomous quality-control pipeline ensures consistency, output is produced via robotic drawing, print, and digital
  • Fusion of cutting-edge technology and handcrafted approach for a luxury brand

Generative Art Practice — Custom Tools & Exhibited Work

An international generative art practice built on custom-developed creative tools — from algorithm editors to autonomous production pipelines.

Custom generative art editor with multiple parameter panels surrounding a canvas of green and pink algorithmic brushstrokes
Artist working with code and generative artwork displayed on a wide monitor — code on left, vibrant abstract output in center
Generative artwork — dense green and white floral algorithmic composition
Role Artist & toolmaker — concept, algorithm design, tool development, production
Stack JavaScript Canvas, WebGL, GLSL shaders, custom editor interfaces
  • Build my own creative tools and editors to develop, iterate, and produce generative artworks
  • Autonomous quality-control pipelines producing output across multiple mediums: robotic drawing, print, web, projection
  • Exhibited internationally: solo shows at DIG Shibuya Tokyo (2025), Code + Matter in Paris (2025); group shows at Bright Moments (Venice, Paris, Berlin), London Biennale, Verse Gallery London, NEORT Shibuya
See artworks → florianzumbrunn.com

Other Projects

Speech-to-Text Pipeline

Multi-model system orchestrating Whisper (GPU via Modal) + Pyannote for speaker diarization. Built as a self-hosted alternative to commercial transcription services.

Goal is exploration and build projects on top of it.

Conots — AI-Powered Journaling

A mind-mapping and journaling tool where AI analyzes accumulated entries over time to surface patterns and extract actionable insights.

Artist Catalog Platform

A tool for artists to create targeted, professional portfolio links for different audiences — galleries, collectors, press.

AI Image Generation Workflows

Commercial prompt-engineering pipelines (Krea, Nano Banana 2) translating complex visual briefs into deterministic, reproducible AI-generated imagery for advertising clients.