Odysseas D
- Research Program Mentor
PhD candidate at EPFL - EPF Lausanne
Expertise
Machine learning and maths and finance
Bio
As a graduate of Cornell University with a deep-rooted passion for artificial intelligence, I’ve dedicated my academic and professional journey to exploring the frontiers of innovation and technology. From conducting rigorous research during my studies to leading impactful projects, I’ve continually sought to bridge theory with real-world application. Currently, I work as an AI Scientist at Hewlett-Packard (HP), where I focus on advancing AI-driven solutions that make a tangible difference. Collaborating with cross-functional teams, I contribute to the development of cutting-edge technologies—always aiming to push the boundaries of what AI can achieve. Outside of work, I’m a passionate soccer player. The sport not only keeps me active, but also sharpens my strategic thinking and teamwork—skills that seamlessly translate into my professional life. Whether it’s on the field or in the lab, I thrive on challenge, collaboration, and continuous growth.Project ideas
Stock price prediction
Imagine having a powerful tool at your fingertips that can analyze massive amounts of financial data, market trends, and historical patterns to predict stock prices and identify the best stocks to buy. This system leverages cutting-edge AI and machine learning technologies to provide data-driven insights, transforming the often unpredictable world of stock trading into a strategic, informed process. By partnering with me, you'll gain access to this advanced technology, which not only enhances the accuracy of stock predictions but also helps mitigate risks and maximize returns, giving you a distinct advantage in the market. Collaborating on this project means being part of an innovative approach to stock trading that saves time and increases efficiency. Instead of spending countless hours researching and monitoring the market, you can rely on our system to do the heavy lifting, providing you with actionable insights that help you make smarter investment decisions. This is more than just a tool; it's a game-changer that can significantly improve your trading outcomes and position you for long-term success in the financial markets.
Medical Chatbot
Imagine having a trusted medical advisor available 24/7, right at your fingertips. A medical chatbot provides exactly that—a convenient, reliable source of health information and guidance that users can access anytime, anywhere. This chatbot can assist with a range of tasks, from symptom checking and triage to offering advice on minor ailments and providing up-to-date information on various health topics. By leveraging advanced AI and a comprehensive medical database, the chatbot can offer immediate responses, helping users make informed decisions about their health, whether it’s managing symptoms at home or determining the need for professional care. Collaborating on this project means contributing to a tool that can significantly improve access to healthcare, especially for those in remote areas or with busy schedules. The chatbot can help reduce unnecessary visits to healthcare providers, easing the burden on medical systems while ensuring users get the support they need. Additionally, it offers a confidential platform for discussing sensitive health issues, empowering users to seek advice without fear of judgment. This medical chatbot isn't just a technological innovation; it's a step towards more accessible, efficient, and user-friendly healthcare.
🌟 Project Title: “GameSense: Build Your Own AI Game Opponent!”
GameSense is an interactive project where students can build and train their own AI-powered game enemy. It starts out as a beginner, but learns from how you play and gets better every time you battle it. Think of it like creating a mini video game boss that evolves with you. 🎮 What can students do? Play a simple online game (like dodge-the-enemy or capture-the-flag). Customize the enemy’s brain using sliders or beginner-friendly code. Train the enemy AI by letting it play against you—or by giving it feedback. Watch how it improves over time, using real AI! 🧠 What’s the AI part? It uses Reinforcement Learning—an AI method where the computer learns by trying, failing, and getting better. Every time the enemy plays, it updates its strategy. 💡 Why will students love it? It’s fun and interactive. No boring math upfront—just play and tweak. It shows how games and AI connect. It teaches real AI concepts in a hands-on way. You can share your trained AI bot with friends or on social media. 🛠 Tech Stack (Behind the Scenes): Frontend: HTML/CSS/JavaScript (or Unity WebGL) Backend: Python (Flask or FastAPI) AI: Stable-Baselines3, PyTorch, or Unity ML-Agents Deployment: Host on Heroku, GitHub Pages, or Replit
QDigit: Can a Quantum Computer Read Handwritten Numbers
🔍 What is it? QDigit is a fun and interactive web project where students can see how a quantum computer tries to recognize handwritten numbers (like 0–9). It’s like teaching a futuristic robot to understand your handwriting using quantum physics and AI. 🧠 What’s the AI + Quantum part? It uses a Quantum Neural Network (QNN)—a type of AI model that runs on quantum circuits instead of normal computers. You feed it simple images (like digits from the MNIST dataset). The quantum computer learns patterns and predicts what number it’s seeing. 🎓 Why is it exciting for students? It makes quantum computing feel real and accessible, not just theory. You interact with real machine learning, in a brand-new way. It’s a conversation starter: "I trained an AI... on a quantum computer!" Great for students curious about physics, AI, and the future of computing. 🖥️ Website Features: 🎨 Draw a digit on-screen (like with a mouse or finger). 🚀 Send to Quantum AI – see what number it guesses. 📊 Compare with Classical AI – which one does better? 🔍 Learn Mode: Explains how quantum bits (qubits) and gates work. 🧪 Test Mode: Try different settings and see how the AI responds. 🛠 Tech Stack (Behind the Scenes): Frontend: HTML/CSS/JS or Streamlit (easy interactive app) Quantum AI: Use PennyLane (by Xanadu) or Qiskit (by IBM) Dataset: Mini MNIST or hand-drawn digit input Backend: Python (to simulate quantum circuits or send to real quantum hardware via API) Deployment: Hugging Face Spaces, Streamlit Cloud, or Replit 🔗 Bonus Engagement Ideas: Leaderboard: See whose digit was most confusing to the AI! “Quantum vs Classic” Battle: Try the same input on both models. Try Real Quantum Hardware: Button that sends your digit to IBM Q for actual processing.