Odin Moja

Hello and welcome. I am an AI researcher and seasoned software engineer. I am a visiting scholar at the FSU Information Institute. I took a brief hiatus from my software engineering career to pursue and complete my PhD at the Florida State University iSchool. I was advised by Dr Marcia Mardis. I completed my PhD in May of 2026. My Dissertation was titled "The Machine’s Eye Does Not Tire: Experimental Exploration of Synthetic Data in Subject Detection of Florida Manatees". My study focused on the use of diffusion model generated manatee image data to aid data needs for training object detection models and producing more robust models. I am currently interested in roles in academia, research and development (private or public), and software engineering.

If you are interested in my exploits as a software engineer, please check out my background here. I have almost a decade of experience, primarily as full stack C#.Net developer but I have also worked with languages like python and java. I have been blessed to work in a wide array of company sizes and industries but in particular those of Finance, Health, Insurance and property management.

Email  /  CV  /  Software Engineering Experience  /  Scholar  /  Linkedin  /  Github

Research

My current interests and focus lay in the realms of machine learning, computer vision, generative AI, optimisation problems, and synthetic data. I am focused on the use of models in realms of daily impact such as medicine, finance, engineering, agriculture and scientific imaging to name a few. I have broad range of personal interests which often leads to me collaborating as part of a multidisciplinary team. Below is a collection of my conference and journal papers.

Moja, O. O. (2026). The Machine’s Eye Does Not Tire: Experimental Exploration of Synthetic Data in Subject Detection of Florida Manatees [PhD Thesis, The Florida State University].

My PhD dissertation in which I leveraged Flux.2 generative AI with a LoRA to supplement and augment genuine training data for the YOLO object detection model. I achieved promising results and synthetic data proved to be useful in aiding the already lightweight, robust performance of YOLO.

Adams, J., Moja, O., & Ko, C. (2026, March 13–14). Comparison of authentic media and synthetic media in computer vision training [Paper presentation]. 28th Annual Southern Association for Information Systems Conference, Asheville, NC, United States. aisnet.org

Air-sea rescue is a highly time sensitive endeavour. Victim submersion time strongly influences survival time. The problem is exasperated by the vast search spaces presented by the ocean. Our study used the combination of several tools to produce an automated synthetic data pipeline which both generated and labelled images. Our results showed the model performed well even in comparison to studies using purely authentic data.

Ravuri, B., Mardis, M. A.,Moja, O., Ayinde, L., & Emdad, F. B. (2026). Complex questions and quality answers: Comparing CHATGPT and Gemini as research collaborators. Journal of the Association for Information Science and Technology, asi.70056. https://doi.org/10.1002/asi.70056

Exploring and comparing the performance of major Large Language Models (LLMs) as research collaborators. Our analyses demonstrated a strong, statistically significant association between question complexity (READ) levels and further assistance.

Adams, J., Roberts, A., Obermaier, L., Ravuri, B., & Moja, O. (2023). Uses of Artificial Intelligence in Higher Education. EdMedia, 1181–1186. https://www.learntechlib.org/p/222633/

This paper investigates the distribution and focal areas of research being conducted in higher education between the years of 2019 and January, 2023. The analysis examined 723 articles resulting from a structured search of the Gale Onefile, ERIC, and Education Full Text databases. The search strategy included peer-reviewed publications published after 2019, filtered for key terms that included artificial intelligence, higher education, and teaching.

Moja, O. O., & Malan, K. M. (2018). Issues and a proposal for automated retirement portfolio risk matching in South Africa. 2018 Conference on Information Communications Technology and Society (ICTAS), 1–6.

This paper is an investigation into the issues related to introducing automation to portfolio optimisation for risk matching.

Miscellanea

Much thanks to Dr. Jon Barron for the template for this page. His repostory has the source code.