Computer Vision Deep Learning R&D Engineer at SECERN AI, Seoul 🇰🇷
I research computer vision models and bring them into production — building and optimizing deep learning models for real-world vision products: face detection & recognition, model compression/quantization, and high-performance inference serving.
- 🏢 Most of my recent work lives on my work account → @syshin-cubox-ai
- 🔬 Interests: Face Detection · Face Recognition · Model Optimization (Quantization) · Inference Serving
- SEEUON — Research & deployment of the core AI models for SEEUON, a remote-work security solution
- Ultralytics contribution — Merged a PR that recovers mAP loss by keeping the DFL in float during OpenVINO INT8 quantization
- GPU infrastructure — Built and operate a GPU server cluster with Slurm-based scheduling (16× H100, 10× A100, 7× A10, etc.)
- Shin, S., Lee, S., & Han, H. (2021). EAR-Net: Efficient Atrous Residual Network for Semantic Segmentation of Street Scenes Based on Deep Learning. Applied Sciences, 11(19), 9119. https://doi.org/10.3390/app11199119
- Shin, S., Han, H., & Lee, S. H. (2025). Improved YOLOv3 with duplex FPN for object detection based on deep learning. International Journal of Electrical Engineering & Education, 62(2), 127–143. https://doi.org/10.1177/0020720920983524
- 📚 Tech notes: Face Detection
Languages
Deep Learning & Computer Vision
Optimization & Serving
MLOps & Infrastructure
