Back to Works
Generative AI
AI Music Video Generation Platform
Solo-built AI music video generation platform that orchestrates external music/video providers, FFmpeg post-processing (composition/subtitles/overlays), and YouTube upload through Celery-based async workflows.

Tech Stack
PythonFastAPIPostgreSQLCeleryRedisReactTypeScriptFFmpegMoviePy
Year
2025.12 - Present
Executive Summary
The Challenge
Long-running external generation jobs required robust orchestration, progress tracking, and retry handling without blocking UX.
The Solution
Built domain-scoped FastAPI routers (projects/songs/videos/youtube/workflows) and a Celery app with retry/time-limit policies. Implemented provider-specific generation tasks, an FFmpeg-based editing service, and a YouTube upload service integrated into the async pipeline with DB-backed status tracking.
Technical Documentation
Methodology & Implementation
Overview
- Solo-built music video generation platform automating generation → post-processing → upload workflows.
Purpose & User Flow
Purpose
- Productize a production pipeline from an idea to a publishable artifact.
User Flow
- Create a project → generate music → generate video → compose/edit (subtitles/overlays) → upload, tracked as step-level statuses.
Key Components
- FastAPI API for job creation/status tracking
- Celery + Redis for long-running generation/post-processing/upload tasks
- FFmpeg/MoviePy for audio/video composition and subtitle/post-processing
Operational Focus
- Failure-first design: status tracking, retries, and idempotency safeguards.