Arabic medical AI · OCR · document intelligence · medical translation
I build practical, testable tools for Arabic and medical document processing, with an emphasis on:
- OCR / handwriting recognition — Arabic medical documents, preprocessing, routing and evaluation.
- Human-in-the-loop training — correction, provenance, datasets and reproducible exports.
- Medical terminology & translation — bilingual glossaries, translation tooling and knowledge bases.
- Linux tooling — safe maintenance and document-processing utilities.
| Project | Role |
|---|---|
| omni-medical-suite | Main medical document intelligence platform |
| ocr-core | Reusable OCR core and engine integrations |
| omni-ocr-training-db | Private OCR training-data store and sync target |
| arabic-medical-glossary | Arabic medical terminology |
| dictionaries-csv | Reproducible bilingual dictionary conversions |
| medical-translation-bot | Medical translation workflow and evaluation |
| translation-core | Reusable translation foundation |
| manjaro-care | Safe Manjaro maintenance tooling |
| channel-ops-dashboard | Channel operations and OCR review tooling |
Evidence before claims · tests before merge · provenance over convenience · fail closed for security-sensitive paths · no silent OCR fallback.
Projects are consolidated when they duplicate a stronger maintained implementation; archived repositories remain available as historical references.
- OCR: preprocessing, multi-engine routing, Arabic RTL handling, evaluation and HITL correction.
- Data: medical glossaries, bilingual dictionaries, ground-truth and training artifacts.
- Translation: reusable core + medical-domain workflows.
- Operations: Telegram/channel tooling and Linux maintenance.
For the current state of the portfolio, start with the repositories above rather than older archived demos.
Built for reproducible Arabic medical AI.