Hi I am Odysseas (pronounced as oh - dh ee s - s EH - aa s). I am currently a PhD student at the University of New Mexico, under the supervision of Prof. Eirini Eleni Tsiropoulou, and a member of the PROTON Lab. My research focuses on leveraging Reinforcement Learning (RL) techniques to solve real-world complex problems. I am particularly interested in creating agents that will be able to tackle tough challenges in fields such as Edge AI, Smart Grid Systems, Electric Vehicle Networks, Multimodal applications, and Natural Language Processing (NLP) applications.
My prior studies include a MSc, a BSc and a Minor degree. More specifically, I completed my MSc and my BSc in Computer Science at the Athens University of Economics and Business (AUEB) under the supervision of Prof. George C. Polyzos and my Minor in Philosophy at Deree - The American College of Greece. I have also worked as an Applied Machine Learning Researcher at Helvia.ai, specializing in the creation of chat agents that utilized state-of-the-art NLP techniques and LLM models , and at Plegma Labs, specializing in the creation of RL agents that optimized energy consumption and electricity costs in Smart Homes.
Feel free to reach out to me if you are interested in collaborating or if you have any questions about my research! The time in my area is:
PhD in Computer Science, 2024-present
University of New Mexico
MSc in Computer Science, 2022-2023
Athens University of Economics and Business
BSc in Computer Science, 2018-2022
Athens University of Economics and Business
Minor in Philosophy, 2019-2021
Deree
Responsibilities include:
Responsibilities include:
The project energy Optimization of building Internet Of Things Infrastructures in a Stratified way presented a holistic approach for an AI-enabled EMS for building-level energy management and max- imizing renewable energy sources utilization. The solution integrated a Photovoltaics (PV) generation forecasting module, a building energy demand forecasting module, and an energy task scheduling and optimization component. This paper presents the methodology and preliminary results regarding the PV generation forecasting and energy task scheduling modules tested for the premises of Institute Mihajlo Pupin.
Standard Full-Data classifiers in NLP demand thousands of labeled examples, which is impractical in data-limited domains. Fewshot methods offer an alternative, utilizing contrastive learning techniques that can be effective with as little as 20 examples per class. Similarly, Large Language Models (LLMs) like GPT-4 can perform effectively with just 1-5 examples per class. However, the performance-cost trade-offs of these methods remain underexplored, a critical concern for budget-limited organizations. Our work addresses this gap by studying the aforementioned approaches over the Banking77 financial intent detection dataset, including the evaluation of cutting-edge LLMs by OpenAI, Cohere, and Anthropic in a comprehensive set of few-shot scenarios. We complete the picture with two additional methods. First, a cost-effective querying method for LLMs based on retrieval-augmented generation (RAG), able to reduce operational costs multiple times compared to classic few-shot approaches, and second, a data augmentation method using GPT-4, able to improve performance in data-limited scenarios. Finally, to inspire future research, we provide a human expert’s curated subset of Banking77, along with extensive error analysis.
Please do not hesitate to contact me!