It starts with defining the research or business problem

My passion is helping clients with data projects that have impact.

Projects start with a good idea. A proof of concept (PoC) shows if the idea will work. From that basis, continuous iteration will allow to improve the project to the desired stage.

In collaboration with different teams, I have successfully managed projects from A-Z. This includes computer vision projects, such as object detection, image classification, as well as e-commerce data analyses.

Breaking down the question in different steps

All projects can be broken down into different building blocks. Often, it includes one or more of the following steps:

Implementation of a database
Preparation of cloud or local environment
Data collection and annotation
Data exploration and visualization
Data preprocessing and quality inspection
Training and validation of machine / deep learning models
Deployment of models
Monitoring of model performance in production

Examples of applications

Machine / deep learning can be used to classify certain characteristics of a photo without human intervention. For example, it can be used to identify the principal color of a photo, which can be used as tags in aggregated e-commerce sites.

Object detection is another area of applications. Please see below how it serves in spotting product defects.