Multiple Myeloma is a type of blood cancer of plasma cells that can severely damage the bones and kidneys and can lead to death. For early-stage detection of this cancer, classification and segmentation of multiple myeloma cells via Computer-Aided Diagnosis (CADs) could be very helpful. We used Seg-PC 2021 challenge dataset consisting of microscopic images of stained cells, and their ground truth masks drawn for cancerous cells in the image by experts. Classification of cancerous and non-cancerous cells was performed using Convolutional Neural Networks along with Transfer Learning from several pre-trained Neural Networks as feature extractors, followed by a trainable dense classification layer. Our classification model achieved an accuracy of 97% on pre-trained MobileNet model training. For segmenting the cancerous cells into nucleus and cytoplasm, we used U-Net architecture, and achieved good performance with mean intersection over union (mIoU) of 86.03%.