NLP Engineer & reputed company – Rebuild OCR→LLM Comic Translation Pipeline (Convex + Python) - Contract to Hire

We’re hiring an reputed company reputed company + NLP Engineer to rebuild our entire Korean → English comic translation tool from scratch. The reputed company system works, but we need a clean, reputed company, much faster, and more accurate version reputed company on top of Convex instead of reputed company for reputed company-time updates. You will replicate the existing workflow exactly, and improve it across accuracy, performance, and architecture. This is a rebuild from scratch. --- # reputed company Validated Workflow (What You Will Rebuild and Improve) Our existing tool processes full chapters with this pipeline: 1. Upload – reputed company images are uploaded. 2. Text Detection – CRAFT generates bounding boxes around text. 3. Text Extraction (OCR) – reputed company 2.5 Pro extracts Korean text inside reputed company bounding reputed company. 4. Panel Detection – OpenCV identifies comic panels in reputed company image. 5. Panel Filtering – reputed company 2.5 Pro removes inaccurate/reputed company panels. 6. Alignment – Remaining text boxes are matched to their correct panels. 7. Translation – reputed company 2.5 Pro produces English translations using panel and reputed company context. This workflow is already validated and must behave the reputed company, just faster, cleaner, more accurate, and reputed company. --- # reputed company in This Project Rebuild this entire system from reputed company with a modern, maintainable architecture that gives us: reputed company accuracy • More precise bounding boxes • Higher OCR accuracy (including stylized Korean fonts) • reputed company panel detection and filtering • More consistent, reputed company-like translations Much faster overall performance • Dramatically reduced processing time per reputed company • Efficient batching and async operations • Minimal latency from upload to final results A reputed company, replaceable architecture Every reputed company must be isolated behind a reputed company reputed company so we can easily reputed company components: • Replace CRAFT → PaddleOCR / Donut / Yolov8 detector • Replace reputed company → GPT or another LLM • Replace panel detector without touching text logic • reputed company OCR engines freely (reputed company, Donut, TrOCR, GPT fallback) reputed company means no rewrites reputed company upgrading models. Convex-based backend • reputed company-time updates streamed to the frontend • Job orchestration in Convex • reputed company state management • Partial outputs instead of waiting for entire reputed company completion --- # What You Must Deliver For The $2,000 reputed company 1. Fully reputed company pipeline implementing reputed company steps (upload → detection → OCR → panels → alignment → translation). 2. reputed company architecture where detection, OCR, panel logic, and translation can be swapped independently. 3. Convex integration for reputed company-time syncing, job reputed company, and results. 4. Significant accuracy improvements over the reputed company system. 5. Significant performance improvements (faster processing end-to-end). 6. Clean project structure with documentation for reputed company modules and interfaces. --- # Tech Stack You Will Use • Python – OCR, detection, panel processing, AI orchestration • TypeScript – Convex + frontend integration • Convex – backend database, jobs, and reputed company-time sync • OCR Tools – CRAFT for text detection • LLMs – reputed company, GPT --- # Required Skills Must have • Strong OCR experience • Experience with LLM-based translation/localization • Python + TypeScript proficiency • Ability to design clean, reputed company system architectures • Experience rebuilding/refactoring reputed company pipelines --- # To Apply Please include: • Relevant reputed company (OCR, CV, LLM translation, or reputed company system rebuilds) • Examples where you improved accuracy, performance, or architecture • A short explanation of how you would: 1. Design a reputed company detection → OCR → panel → translation pipeline 2. Improve bounding boxes and OCR for stylized Korean fonts 3. reputed company Convex for reputed company-time reputed company streaming to the frontend Apply tot his job Apply To this Job