SELECTED WORK
AI OCR Translator
A Chrome extension that captures, recognizes, and translates any on-screen region, with a Vision LLM and an offline fallback.
What changed
- Combines OCR and translation in one step across 32 languages.
- Provides a Tesseract.js offline fallback and environment-specific release artifacts.
One-step translation
The Manifest V3 extension lets users select an on-screen region and translate its text in one step. It is built with React, TypeScript, and WXT.
Online and offline paths
Images can be processed by a Vision LLM through OpenRouter. A Tesseract.js offline fallback supports OCR across 32 languages.
Delivery
The open-source project uses an A/B-gated GitHub Actions pipeline to publish Chrome artifacts for each environment.