We present a comprehensive, interactive and extensible system to assist in debugging and visualization with a focus on the tasks that are often encountered by the practitioners during the various phases of model development. Our system offers the integrated solution with consistent interface for variety of such tasks including data exploration, architecture visualization, model performance analysis, interactive real-time analysis during model training and explaining predictions after the training. Instead offering these capabilities through a custom application, we emphasize on consistency, extensibility and composibility with existing frameworks through interactive interfaces such as Jupyter Notebook which many practitioners are already familiar with. We also offer new capabilities not available in exisitng systems such as on-demand generation of arbitrary data stream from a live training process using lambda expressions enabling the users to generate multiple simultaneous real-time visualizations of a desired information using interactive queries without imposing stop-change-restart cycle on the training processes. We plan to release our system as an open source cross-platform offering to help accelerate model development tasks in the practitioner community.