Flash Cut
Flash Cut is a video generation platform based on text, pictures, and video clips to generate digital avatar videos.
Website:
https://www.boltcut.ai (English)
https://shanjian.tv (Chinese)
1. Team Members
5 members (Backend)
2. Company
Shenzhen Qutui Science & Technology Co., Ltd.
3. Technologies
- Node.js (Typescript)
- Redis
- MongoDB
- RabbitMQ
- ApolloConfig
- WebSocket
4. My Role
When this project started from scratch, I was transitioning from an iOS Engineer to a Backend Engineer. After 2 months, I became the core developer on the project in August 2021 and continuously worked on it until the first version launched in July 2022.
My primary focus was to develop a feature that converts text and images into video content. This involved close collaboration with the video engine team for rendering and generation, as well as working with the Android, iOS, and Frontend engineers for API integration and troubleshooting.
5. Biggest Challenge
The most difficult part was managing a highly complex system with multiple stages for metadata generation and processing. Any error along the pipeline could lead to video generation failures or unexpected final outputs. We needed to maintain clear visibility over each step of the process to diagnose and fix issues quickly.
6. How I Addressed the Complexity
- Clear Diagrams & Documentation: I created many comprehensive diagrams illustrating the video-generation flow for each step. Detailed documentation explained each component’s role and how they interacted.
- Clear Logging & Error Tracking: I implemented more granular logging throughout the pipeline, which made it easier to pinpoint exactly where and why failures occurred.
- Proactive Communication & Collaboration: I encouraged frequent design discussions; by addressing potential pitfalls early, we avoided major design errors that could harm the project.
7. Result
The project was successfully launched and became one of the earliest leading AI platforms for text and image-to-video generation.
It continued to grow and gain recognition after I left the company. In February 2025, it was featured on Hong Kong Phoenix TV’s list of 8 Chinese AI app recommendations—a list that also included DeepSeek.
Here is the Chinese Video link from Phoenix TV