About

Aditya Bhandari is a Graduate Analyst at Barclays UK with 2.43 years of relevant experience in the field of Software Engineering. He is currently designated as a Senior Software Developer in the Market Surveillance team of the Investment Banking Division at Barclays. Aditya has worked on four ongoing projects, namely the Honda project, Magicbricks project, Fraud Detection COE, and Customer Segmentation COE, where he has showcased his expertise in Python, Pandas, Web Development, OpenCV, CNN, Deep Learning, and Tensorflow. In the Honda project, Aditya preprocessed and visualized a 9 million automotive dataset using Pandas and ArcGisPro software to find the most accident-prone areas on Delhi streets. He also performed time series analysis to find the severity of accidents over hours of the day after ANOVA. In the MagicBricks project, he hypothesized various algorithms to merge four different datasets for three months, creating a 2.2 million master dataset to determine factors like location, call hour, etc. affecting the lead generation and retention of real estate properties for Magicbricks. Aditya also trained a 3 layered (14-7-14 neuron) Autoencoder model to detect fraudulent bank transactions using the Anomaly Detection Technique in the Fraud Detection COE project. He developed an interactive web analytical application using Dash Framework to build and visualize the model. In the Customer Segmentation COE project, he successfully implemented the PAM algorithm (k-medoids) on the categorically featured customer dataset to find clusters for customer segmentation. He developed an interactive dashboard using Dash (Python) to visualize the clusters. Aditya has also worked as an Internship (Data Analyst) at NRI and as a Summer Project (Remote) at Osaka University. At Osaka University, he worked on detecting Sign Language Alphabets from real-time video feed using Python, OpenCV, Tensorflow, and MNIST Sign dataset. He trained a CNN model using Dropout for regularization, softmax as activation function, and SGD optimizer for compiling the model. He also created a window using OpenCV to get the required input from the webcam and transformed it to a 28x28 skinmasked grayscale image. For feature extraction, he used the Speeded Up Robust Feature (SURF) technique to extract descriptors from the segmented gesture image. Aditya completed his education from IIT, where he pursued his Bachelor's and Master's degree in Software Engineering