Computer Vision/ML Engineer -PT Freelance - Worldwide

Job Description:

Care AI is developing cutting-edge technologies that have real-world applications in areas such as healthcare, surveillance, and human-computer interaction. We rely on computer vision for image classification and video based action recognition. We are looking for an engineer to develop high-performance vision models that operate reliably in real-world settings. If you’re passionate about building AI that sees and understands the world, this role is for you.

In this role, you will:

• Develop and implement computer vision algorithms for object detection, image segmentation, image classification and action recognition from video streams
• Process and analyze video data including data cleaning, augmentation, and preparation for model training.
• Develop and test deep learning models, such as 3D Convolutional Neural Networks (3D CNNs) or Vision Transformer for video classification
• Analyze model performance and contribute to the optimization of algorithms for improved accuracy and efficiency.
• Work with image and video datasets to drive AI improvement
• Write robust production code for deploying the above algorithms on hardware platforms.

Qualifications:

• BS, MS, or PhD in AI, Computer Science, or a related field, or at least 3-4 years of industry experience
• Strong background in machine learning algorithms and concepts in general.
• 2+ years experience developing high-performance deep learning models for vision
• Proficiency in Python and deep learning libraries like openCV, OpenPose, PyTorch or TensorFlow. Proficiency in C++ is a plus.
• Knowledge of ML-Ops fundamentals (release definitions, staging/test environments, cloud deployments).

Nice to Have:

• Familiarity with datasets commonly used in action recognition research, such as UCF101, HMDB51, or Kinetics.
• Knowledge of state-of-the-art architectures for video analysis, including two-stream networks and temporal modeling approaches.
• Experience with data annotation tools and techniques for video datasets.