Current Vacancies

Research Assistant (3D Vision and Multimodal Alignment)

研究助理（三维视觉与多模态对齐方向）

岗位职责

1.负责多模态影像的跨域表征学习、图像到图像翻译（Image-to-Image Translation）及生成式桥接算法研发。

2.负责基于深度学习的图像特征匹配、跨模态检索与三维空间姿态估计（6-DoF Pose Estimation）算法研发。

3.负责多维影像空间对齐、几何变换预测等算法的工程实现、端到端流水线搭建与性能迭代。

4.协助团队撰写高质量技术报告、专利申请书以及国际顶会/顶刊学术论文。

任职要求

1.专业背景：计算机、自动化、测绘遥感、机械工程、生物医学工程或相关专业，本科及以上学历（优秀的应届本科毕业生亦可）。

2.编程与框架：工程能力扎实，精通 Python 编程，熟练使用 PyTorch，具备扎实的线性代数、矩阵运算与三维几何变换数学基础。

3.核心技能：深入理解计算机视觉基础，在图像配准/对齐、图像翻译（GAN/Diffusion）、特征匹配与图像检索、或三维视觉/姿态估计等方向至少有一项扎实的实战经验。

加分项

有跨模态数据对齐或医学影像（如 CT/MRI/超声）处理与三维联动项目经验者优先；

熟悉三维点云处理、经典 3D 几何算法（如 Homography, ICP, PnP）或有限元力学分析者优先；

具备良好的英文文献阅读与理解能力，能够快速复现前沿顶会（如 CVPR, ICCV, MICCAI）算法。

申请方式

请将个人简历发送至 hr02@cair-cas.org.hk，邮件主题请注明：应聘岗位－姓名－官网投递。

Job Responsibilities

Multimodal Alignment & Generation: Develop cross-domain representation learning, image-to-image translation, and generative bridging algorithms for multimodal imaging data.

Retrieval & Spatial Localization: Design and optimize deep learning algorithms for image feature matching, cross-modal retrieval, and 3-D spatial pose estimation (6-DoF Pose Estimation).

Engineering Implementation: Responsible for the engineering execution, end-to-end pipeline setup, and performance iteration of multidimensional image alignment and geometric transformation prediction.

Academic Output: Assist the team in drafting high-quality technical reports, patent applications, and research papers for top-tier international conferences/journals.

Job Requirements

Educational Background: Bachelor’s degree or higher in Computer Science, Automation, Remote Sensing, Mechanical Engineering, Biomedical Engineering, or related fields. (Outstanding recent undergraduates are strongly encouraged to apply).

Programming & Frameworks: Solid engineering skills with proficiency in Python. Deep hands-on experience with PyTorch. Strong mathematical foundation in linear algebra, matrix operations, and 3D geometric transformations.

Core Competencies: Deep understanding of computer vision fundamentals. Proven practical experience in at least one of the following domains: Image Registration/Alignment, Image-to-Image Translation (GAN/Diffusion), Feature Matching & Retrieval, or 3D Vision/Pose Estimation.

Preferred Qualifications

Prior experience in cross-modal data alignment or medical imaging (e.g., CT/MRI/Ultrasound) processing and 3D reconstruction projects is highly preferred.

Familiarity with 3D point cloud processing, classic 3D geometry algorithms (e.g., Homography, ICP, PnP), or finite element method (FEM) analysis is a plus.

Strong English literacy with a proven track rate of quickly reproducing algorithms from top-tier conference papers.

Application Method

Please send your resume to hr02@cair-cas.org.hk. For the email subject line, please indicate: Application for [RA - 3D Vision and Multimodal Alignment] - [Name] - [Applied via CAIR Official Website].