Lessons from managing the Growth Hackers project to build Langflix's Shorts recommendation system.
Lessons from improving Shorts recommendations, from analysis and data pipelines to A/B test design.
The end-to-end development of Langflix's recommendation system, from behavioral signals and Two-Tower training to cold-item evaluation and real-time TypeScript serving.
How we grouped more than 8,000 Shorts by the meaning of their content rather than by manually assigned tags.
A look back at the analysis and decisions behind the Langflix Shorts recommendation system built by Theta One and Growth Hackers at Seoul National University.
Theta One AI presented its Korean-English code-switching speech recognition evaluation framework, HiKE, at EACL 2026.
Discover how to learn English with your favorite YouTube and Netflix content!
Even a multilingual speech recognition model can struggle when Korean and English are mixed in the same utterance.
Theta One AI presented a paper at ACL 2025 in Austria and explored current trends in AI research.
