Hi,
I am an engineer who is currently employed in the Machine Learning team in a tech company. I mostly work on computer vision and deep learning projects and I work with CNN and other ML algorithms daily.
I have worked on various CNN architectures like Alexnet, Inception, VGG, Resnet, Desnenet, etc, and transfer learning, fine-tuning them for various classification and comparison/matching applications. Moreover I have implemented custom loss functions in training several models such as triplet loss for people re-id and face recognition.
I am familiar and have worked with Face detection, recognition and attribute classification of people, and a good background in computer vision as well. It is indeed possible to train a selected set of layers and if needed, fine-tune the entire model and integrate new layers for an existing model.
I have and am working with various deep learning frameworks in python such as Keras (tf backend), tensorflow, pytorch, caffe and familiar with other packages such as opencv, scikit-learn, etc.