WatchTower
A public reconstruction of an internship project that evaluated Korean-to-English product copy with multiple quality signals and post-edited only translations that required review.
- Prioritized translations likely to contain errors instead of applying the same review depth to every sentence.
- The internship evaluation measured up to about 2,500 sentences per hour and a 68% quality-grade improvement among post-edited translations.
Python · LaBSE · COMET · LLM Evaluation · NLP




