💡 Specializing in Vision + Language, building complete solutions for document AI, face recognition, and eKYC.
I am a Senior Computer Vision & Machine Learning Engineer, with deep expertise in both vision and language domains.
My key strengths include end-to-end development, infrastructure design, and cross-functional deployment.
- Specializing in Document OCR/KIE, Face Recognition, and Face Anti-Spoofing Detection
- Experienced in model design and training, including CNN, Transformer, ViT, and self-supervised learning
- Proficient in MLOps workflows, including ONNX Runtime, quantization (PTQ/QAT), Docker, and CI/CD
- Self-hosting a full-stack technical blog and research note platform using React, FastAPI, and Nginx
- Demonstrates strong capabilities in Linux system administration and DevOps practices
📂 Most of my professional work is maintained under the DocsaidLab organization.
This profile contains only personal repositories and prototypes.
| Stack | Proficiency | Notes |
|---|---|---|
| Python / PyTorch | Expert | Model training, CNN, Transformer, ViT, SSL |
| Computer Vision & Deep Learning | Expert | OCR, KIE, Face Recognition, Anti-Spoofing |
| ONNX Runtime / PTQ / QAT | Proficient | Inference optimization & quantization |
| Docker / Docker Compose | Proficient | Containerization & private deployment |
| FastAPI / RESTful API | Proficient | Lightweight backend & eKYC integration |
| Nginx / Linux DevOps | Proficient | HTTPS, reverse proxy, CI/CD automation |
| Docusaurus (React) | Proficient | Multilingual docs & blog framework |
- DocAligner: Detects document corners for alignment.
- DocClassifier: Document image classification pipeline.
- MRZScanner: Finds MRZ regions on ID/passport images.
- AutoTraderX: Integration experiments with Taiwan brokerage APIs.
- Capybara: CV toolkit for batch inference and utilities.
- GmailSummary: Gmail ⇄ OpenAI summarizer (archived).
- Nginx Notes: Handy Nginx configs & tips.
- WordCanvas: Synthetic font‑to‑image data generator.
Covers topics like ViT, OCR, Multimodal Learning, Self-Supervised Methods, and more.
- 🧾 Browse all notes: https://docsaid.org/en/papers/intro
👍 Follow me on Facebook Fan Page for updates on computer vision, AI papers, and project insights.