aspiring AI engineer

Syeda zainab  • Peerlist

This project focuses on predicting customer churn for a bank by classifying whether a customer will leave the bank (1) or stay (0). The main goal was to build a beginner-friendly deep learning workflow and understand how Keras and TensorFlow operate behind the scenes. I worked in google colab using kaggle  inbuilt Keras interface, along with Pandas, NumPy, and Matplotlib for data preprocessing, analysis, and visualization. I developed two versions of the model: the first demonstrates the basic components of building a neural network, and the second includes improvements such as accuracy evaluation, validation split, threshold adjustments, and tuning epochs to enhance model performance. Through this project, I strengthened my understanding of neural network layers, training behavior, and overall deep learning workflow.

Customer Churn Prediction - Deep learning Pro

A project where we started our learning of Machine Learning and implementing it to a biological problem known as Coronary Artery Disease (CAD). We first applied ML algorithms to a cardiovascular risk factor dataset and predicted the risk factors, after that we implemented Navier-Stokes Equations to a neural network algorithm making it Physics-informed neural network (PINNS). We applied PINNS in our data to calculate the Wall Shear Stress (WSS) of the blood in the diseased artery (stenosis).

Machine Learning applied to coronary artery d

This project focuses on classifying handwritten digits (0–9) using the MNIST dataset with a Deep Learning model. The dataset consists of 28x28 pixel grayscale images, which were preprocessed and fed into a neural network. Through this project, I gained hands-on experience with: 1)Deep Learning architectures ,building and understanding the structure of neural networks. 2)Activation functions ,exploring how functions like ReLU and Softmax impact learning and predictions. 3)Model training and evaluation ,splitting data, training the model, and testing accuracy 4)Image data representation ,understanding how pixel values are transformed into meaningful features for the model. The final model was able to accurately recognize handwritten digits, reinforcing my understanding of how Deep Learning models process and classify images. The model achieved 97% accuracy on test data.

MNIST digit classification (DL project)

I recently completed a machine learning project where I built models to classify emails into spam and ham. The workflow included text preprocessing and data cleaning, checking data distribution, and handling class imbalance using SMOTE and resampling techniques. For feature extraction, I applied TF-IDF vectorization to convert raw emails into numerical features. I then trained and tested two models—Naive Bayes and Logistic Regression—achieving accuracies of 96% and 95% respectively. This project not only strengthened my skills in text preprocessing, feature engineering, and model evaluation but also gave me a deeper understanding of techniques that form the foundation for Deep Learning and Generative AI.

Spam classification project

The project is about the Vrinda store (Indian Store) that wants to create an annual report for understanding their customers needs and requirements and grow more in the future. Performed data cleaning, preprocessing and data visualization to make great and actionable insights.