ByteDance VP Hong Dingkun: AI Coding practices and exploration
AI summary
At the Volcano Engine Force conference, ByteDance VP Hong Dingkun shared AI Coding practices. Over the past year, ByteDance's AI code contribution rate grew 6x and token consumption grew 5x, but over-focusing on a single metric can be misleading—over 90% of code in the TRAE team is AI-generated, yet per-capita requirement throughput improved only 60%. 900 experiments show mainstream Coding model combinations exceed 80% code correctness, but deliverability is only 40-60 points; combined with Harness infrastructure, it rises to 80 points. AI lowers the programming barrier but requires optimizing metrics, governance, and collaboration. ByteDance explores prototype-driven development, distills capabilities into TRAE (daily average token consumption 5.6 trillion, up 50x), and launched TRAE Work.
Key Takeaways
- At the Volcano Engine Force conference,ByteDance VP Hong Dingkun shared AI Coding practices.
- Over the past year,ByteDance's AI code contribution rate grew 6x and token consumption grew 5x,but…
- 900 experiments show mainstream Coding model combinations exceed 80% code correctness,but…
- combined with Harness infrastructure,it rises to 80 points.
- AI lowers the programming barrier but requires optimizing metrics,governance,and collaboration.
Quick Facts
洪定坤把字节用 AI Coding 一年踩的坑和实验数据摊开讲,尤其‘过度重视代码贡献率’的反思和 Harness 基建的实证,是所有推 AI 编程的团队必看的复盘。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: ByteDance VP Hong Dingkun: AI Coding practices and exploration. June 24, 2026. https://new.nat6.net/en/news/cmqs68cgx0090slfuumt36jny/