Joseph is a Data Scientist specialised in GIS. But, first and foremost, he is an Engineer who enjoys planning and building projects from the ground up.
He studied Spatial Planning & Development Engineering at the Aristotle University of Thessaloniki, from which he received his integrated M.Eng. in 2020. There, he learned research methodologies and how to deeply analyze and evaluate the living space in the economical, social and environmental dimensions using GIS, Data Science and Machine Learning.
After his graduation he invested his time in learning the ins and outs of various ML algorithms and studied Deep Learning through the lectures and paradigms of popular DL advocates such as Andrew Ng and Andrej Karpathy. His favorite Data Science lecturer, however, is Kilian Weinberger.
Along his career he has learned first hand about deploying large scale image processing and computational algorithms in unix environments. He has successfully solved various Data Science problems in Computer Vision and is proficient with most model architectures and problem formulations for the domain. Lately, he has become increasingly skilled in MLOps engineering and aspires to take on designing and developing full scale ML systems in Computer Vision and Natural Language Processing contexts.
Working on an AI-assisted data fusion project, across three public repositories, msi2slstr, msi2slstr-training and msi2slstr-datagen. This project aims to provide a tool for easily downscaling Sentinel-3 Land Surface Temperature rasters using the spatial information of Sentinel-2 images.
Created Machine-Learning-based Forest and Water Quality monitoring solutions at EarthPulse as a Data Scientist and Python Developer.
Solved Computer Vision based building height modeling problems at RSLab as a Research Assistant and AI professional.
Volunteer Market Analyst at MiCreate, using geoinformatics and web scraping tools to plan expansion areas for campaigns.
Received M.Eng. from School of Spatial Planning and Development & Started coding.
As part of my studies in the School of Spatial Planning & Development I authored the thesis “Land Cover Analysis using OBIA: a case study of Oraiokastro, Chalkidona and Delta” , which I defended in 2020.
The term Object-Based in Image Analysis refers to the scientific notion of analyzing an image in generally meaningful chunks or pieces (i.e. groups of pixels), by first segmenting it. Image segmentation as a preprocessing step in image analysis was an almost natural evolution of the classic pixel-based approach long before Convolutional Neural Networks became accessible.
There are plenty of algorithms...
Drop me an email if you are interested in my work!