Before starting this course, I can say my data science projects were quite chaotic. I had learned many concepts on my own, in a disorganized way. My notebooks were hard to maintain, I didn’t follow a clear structure, and I often jumped from one model to another without proper validation or traceability.
Thanks to this course, I was able to organize my thoughts more clearly. The recorded lessons, the hands-on assignments for each topic, and the final projects were all valuable tools that helped me learn how to develop and deploy models in a much more professional way. MLZoomcamp helped me structure my workflow and gain key tools that I hadn’t known how to use correctly before.
Some of the skills I developed include preparing data systematically, applying regression and classification techniques with proper validation, choosing the right evaluation metrics, and most importantly, understanding when and why to use a particular model. I also built confidence working with neural networks, applying transfer learning, and making inferences using models served with Flask or even AWS Lambda.
During the course, we were encouraged to work on two final projects and take part in a Kaggle competition. In my case, I chose to focus on the following:
🩸 Predicting cancer in blood cells
Repository: GitHub - Blood_Cell_Cancer_Prediction
🌪️ Classifying disaster tweets
Repository: GitHub - NLP_with_Disaster_Tweets.
I also wrote an article about this project, where I explore how I used spaCy, TensorFlow, and BERT-based architectures for natural language processing.
I found both projects especially engaging because they deal with real-world challenges. Their complexity allowed me not only to apply everything I had learned throughout the course but also to push myself further by incorporating new concepts that went beyond the core curriculum. Below is a summary of one of them:
Acute lymphoblastic leukemia (ALL) is an aggressive type of cancer that affects the blood and bone marrow, especially in children. Its diagnosis relies heavily on the microscopic analysis of blood cells, a complex, subjective, and time-consuming task for healthcare professionals.
The goal of my project was to develop an artificial intelligence-based solution capable of automatically classifying and segmenting blood cells from microscopic images, combining deep learning models, computer vision techniques, and an accessible user interface.
I’ve always been drawn to the application of data science in areas related to human health. The idea of building a system that can support fast, efficient, and accurate diagnosis and that such a diagnosis could potentially help someone anywhere in the world is my biggest motivation.
In many regions, access to trained pathologists and well-equipped laboratories is limited. In this context, intelligent systems can assist in the early and reliable diagnosis of diseases like leukemia. This project aims to: