ESRA ALIOGLU

Experience

• 

  NHS

  Nov 2022 - Mar 2023

  Royal Voluntary Service

  I actively joined the vaccination program after the Covid-19 period.

• 

  Oxfam

  Nov 2023 - Feb 2024

  Trainee Lead Volunteer

  Successfully fulfilled the roles of both a retailer and a research assistant.Contributed to determining competitive pricing strategies for used books and CDs. Conducted research on market trends and analyzed comparable products to inform pricing decisions.Significantly improved communication skills through interactions with customers and fellow volunteers.

• 

  London South Bank University

  Apr 2024 - now

  I am actively involved in facilitating Spike Prime robotic teaching sessions specifically designed for secondary school students, where I lead hands-on learning experiences to introduce them to the fundamentals of robotics and programming engagingly and educationally. I take on the responsibility of leading and training the ambassadors within the CSI (Computer Science and Informatics) department, ensuring that they are well prepared to represent the department effectively and assist with various outreach activities. I am responsible for organizing and managing events, including university open days, where I coordinate different aspects of the events to ensure a smooth and successful experience for prospective students and attendees. Show less This project revolves around defining breast cancer imaging models using various modelling techniques to perform the most accurate result in breast cancer diagnoses. This project focuses on analysing the GNN, YOLOv8, and U-Net models' imaging outcomes to define the most accurate and sensitive for early diagnosis to reduce the death rate.The BUSI breast cancer dataset is categorized into benign, malignant, and normal images, and each category comprises images along with corresponding mask images. UNet - The stages featured displaying the actual image, image mask, and blended image following the model's definition and subsequent testing/evaluation of the results. Blended images can help in understanding the strengths and weaknesses of the model by providing a clear visual representation of True Positives, False Positives and False Negatives. This model consists of layers, including the initial convolutional layers, down-sampling layers, and up-sampling layers.YOLOV8 - The YOLOv8 model represents the most recent advancement in YOLO technology, providing advanced capabilities for tasks such as object detection, image classification, and instance segmentation. The object image detection method has been used to achieve the most accurate results in terms of tumour detection. It is essential to configure the YAML file for the Yolov8 model, as it allows us to define the locations of the training and validation data.The GNN model insists splitting the dataset for training and testing sets. Image transformation is required to prepare image data for training the model to ensure the images are consistent in size and format for better performance. Train size is 70% of the total number of samples in graph_data while test size is the remaining 30% of the samples. The analysis results showed that the U-Net model achieved the highest accuracy at 0.89, followed by the GNN model with an accuracy of 0.62, while the YOLOv8 model performed below 0.50. Show less

  • Student Ambassador

    Jan 2022 - now

  • Research Assistant

    Apr 2024 - Jul 2024