ML Team Lead

[Client Description] 是上市企業飛捷科技投資的子公司，也是間獨立營運的新創。團隊雖小，但不惜重本打造一支夢幻隊。技術團員分別來自: Microsoft、Appier、HTC 等大型跨國科技企業與 AI Labs、Umbo CV、Gogolook、Cubo AI 等知名新創。團隊包含了深度學習、電腦影像、演算法、前後台軟體設計以及硬體設計等各領域頂尖專家 。

主要客戶包含全球第 2 與第 3 大漢堡連鎖業者 Burger King 和 Wendy’s，以及全球第 3 大披薩連鎖業者 Little Caesars。於 2022 年開始與多家美國客戶展開合作，試營運獲得肯定，並於 2023 年擴大合作，將產品導入更多分店。2025 年更與美國品牌 Zaxby’s 簽署全品牌導入合約，預計導入近千家門市，顯示我們產品在美國市場得到客戶高度信任，且已擁有成熟的技術能力。

[Candidate Profile]

• 7+ years of experience in Machine Learning, Deep Learning, or Computer Vision, with proven experience deploying CV solutions in production environments.
• Experience leading an ML/CV team or driving technical strategy and delivery for ML-based products.
• Strong expertise in computer vision, such as object detection, multi-object tracking, Re-ID, image segmentation, or video analytics.
• Solid understanding of traditional computer vision and image processing techniques, including feature matching, geometric transformations, and image warping.
• Ability to analyze model performance in challenging environments, including low light, occlusion, motion blur, and varying camera conditions.
• Hands-on experience across the ML lifecycle, from dataset preparation and model training to deployment, monitoring, and optimization.
• Proficiency in Python and PyTorch or TensorFlow, with strong capabilities in model evaluation, error analysis, and experimentation.
• Strong problem-solving and communication skills, with the ability to translate business requirements into technical solutions while remaining hands-on.

[Responsibilities]

• Lead and mentor the ML/CV engineering team, overseeing technical development, project execution, and team growth while remaining hands-on.
• Develop and improve computer vision solutions, including object detection, tracking, Re-ID, camera calibration, and event recognition.
• Collaborate with Product and cross-functional teams to translate business needs into technical requirements, measurable goals, and development plans.
• Manage the end-to-end machine learning workflow, from data preparation and model training to evaluation, deployment, and continuous improvement.
• Enhance model robustness and accuracy under challenging real-world environments and varying camera conditions.
• Optimize models for edge devices, balancing accuracy, inference speed, resource utilization, and system stability.
• Develop scalable tools and automated workflows to improve data processing, camera setup, model validation, and deployment efficiency.