February 18, 2026

Overview

MishMash is an AI music project published on Spotify, YouTube, and all major streaming platforms through DistroKid. What makes it a technical project is the fully automated production pipeline built from scratch: from AI music generation to video rendering to multi-platform distribution.

What I Built

• AI music generation pipeline: Using ACE-Step 1.5 with ComfyUI for generating full songs from lyrics and parameters
• Automated video rendering: Custom Remotion (React) projects that generate music videos from background images with synchronized lyrics, transitions, and Ken Burns effects
• Multi-format output: Album videos, single videos, and audiobook chapters all rendered through the same pipeline
• DistroKid integration: Waterfall release strategy with ISRC code management across platforms

Production Output

• His Side album: 11 tracks, fully rendered and published on YouTube
• 6 standalone singles: Each with unique art style (watercolor, paper craft, charcoal, polaroid collage)
• 25 tropical house singles: Written with full ACE-Step parameters
• 100 additional tracks written for KRAEL (solo project) across 40+ genres
• Ember EP: 5-track lofi jazz EP (first letters spell XHULI)

Tech Stack

• ACE-Step 1.5 (ComfyUI) for AI music generation
• Remotion (React/TypeScript) for automated video rendering
• Grok AI for visual asset generation
• DistroKid for music distribution
• Kokoro TTS for audiobook narration

Why This Matters

This isn't just a music project. It's a demonstration of building automated creative pipelines: taking AI tools and stitching them together into a repeatable, scalable production system. The same engineering principles that apply to backend systems (pipelines, queues, batch processing) apply here.