AI Implementation
AI Content Automation
I built an AI pipeline — improved over five versions — that turns a news article into a finished 9:16 TikTok/Reels video: Vietnamese voice-over, word-by-word karaoke subtitles in sync with the audio, matching B-roll and correct brand icons, smooth transitions, and high-quality rendering.
Year :
2026
Industry :
Media / Content Creation
Client :
Personal
Project Duration :
2026
Problem :
Making one good short video by hand takes hours. You write the script, record a voice, time the subtitles word by word, find images and clips, edit the transitions, and render the file. For anyone who needs to publish often — a TikTok channel, a Reels page, or a clinic's marketing — doing this by hand simply does not scale. The work is also repetitive, which is exactly where AI helps.
Solution :
I built a full AI pipeline that turns a news article into a finished 9:16 video. The steps run end to end: article → script → Vietnamese voice-over → subtitles synced to each word → matching B-roll, brand-correct icons, and backgrounds → smooth, professional transitions → a high-bitrate render. I improved it across five versions (v1 to v5), adding hook-and-retention logic so viewers keep watching, shader-based scene transitions, moving text, shared asset libraries, and voice audio mastering. A rule throughout: every video gets a fresh design, never a reused template, so it looks made by hand.
Challenge :
The hard parts were technical. The animation has to be repeatable and “seek-safe,” or the render breaks. The Vietnamese subtitles have to line up word by word with the voice. And each video must look new, not templated, while staying clean and easy to read, not busy. Each version of the pipeline traded a bit more setup for a lot more automation and polish, and getting that balance right took many iterations.
Summary :
The pipeline turns hours of manual video work into a repeatable system. It directly proves one of the tasks in this job — creating marketing content, including TikTok/Reels scripts, blog posts, and emails, with AI. It also connects my 1B-view creator background to real AI engineering: I don't just make content, I built the machine that makes it — which is exactly the mindset a clinic wants for its marketing channel.
More Projects
AI Implementation
AI Content Automation
I built an AI pipeline — improved over five versions — that turns a news article into a finished 9:16 TikTok/Reels video: Vietnamese voice-over, word-by-word karaoke subtitles in sync with the audio, matching B-roll and correct brand icons, smooth transitions, and high-quality rendering.
Year :
2026
Industry :
Media / Content Creation
Client :
Personal
Project Duration :
2026
Problem :
Making one good short video by hand takes hours. You write the script, record a voice, time the subtitles word by word, find images and clips, edit the transitions, and render the file. For anyone who needs to publish often — a TikTok channel, a Reels page, or a clinic's marketing — doing this by hand simply does not scale. The work is also repetitive, which is exactly where AI helps.
Solution :
I built a full AI pipeline that turns a news article into a finished 9:16 video. The steps run end to end: article → script → Vietnamese voice-over → subtitles synced to each word → matching B-roll, brand-correct icons, and backgrounds → smooth, professional transitions → a high-bitrate render. I improved it across five versions (v1 to v5), adding hook-and-retention logic so viewers keep watching, shader-based scene transitions, moving text, shared asset libraries, and voice audio mastering. A rule throughout: every video gets a fresh design, never a reused template, so it looks made by hand.
Challenge :
The hard parts were technical. The animation has to be repeatable and “seek-safe,” or the render breaks. The Vietnamese subtitles have to line up word by word with the voice. And each video must look new, not templated, while staying clean and easy to read, not busy. Each version of the pipeline traded a bit more setup for a lot more automation and polish, and getting that balance right took many iterations.
Summary :
The pipeline turns hours of manual video work into a repeatable system. It directly proves one of the tasks in this job — creating marketing content, including TikTok/Reels scripts, blog posts, and emails, with AI. It also connects my 1B-view creator background to real AI engineering: I don't just make content, I built the machine that makes it — which is exactly the mindset a clinic wants for its marketing channel.
More Projects
AI Implementation
AI Content Automation
I built an AI pipeline — improved over five versions — that turns a news article into a finished 9:16 TikTok/Reels video: Vietnamese voice-over, word-by-word karaoke subtitles in sync with the audio, matching B-roll and correct brand icons, smooth transitions, and high-quality rendering.
Year :
2026
Industry :
Media / Content Creation
Client :
Personal
Project Duration :
2026
Problem :
Making one good short video by hand takes hours. You write the script, record a voice, time the subtitles word by word, find images and clips, edit the transitions, and render the file. For anyone who needs to publish often — a TikTok channel, a Reels page, or a clinic's marketing — doing this by hand simply does not scale. The work is also repetitive, which is exactly where AI helps.
Solution :
I built a full AI pipeline that turns a news article into a finished 9:16 video. The steps run end to end: article → script → Vietnamese voice-over → subtitles synced to each word → matching B-roll, brand-correct icons, and backgrounds → smooth, professional transitions → a high-bitrate render. I improved it across five versions (v1 to v5), adding hook-and-retention logic so viewers keep watching, shader-based scene transitions, moving text, shared asset libraries, and voice audio mastering. A rule throughout: every video gets a fresh design, never a reused template, so it looks made by hand.
Challenge :
The hard parts were technical. The animation has to be repeatable and “seek-safe,” or the render breaks. The Vietnamese subtitles have to line up word by word with the voice. And each video must look new, not templated, while staying clean and easy to read, not busy. Each version of the pipeline traded a bit more setup for a lot more automation and polish, and getting that balance right took many iterations.
Summary :
The pipeline turns hours of manual video work into a repeatable system. It directly proves one of the tasks in this job — creating marketing content, including TikTok/Reels scripts, blog posts, and emails, with AI. It also connects my 1B-view creator background to real AI engineering: I don't just make content, I built the machine that makes it — which is exactly the mindset a clinic wants for its marketing channel.