Google Summer of Code 2025 Proposal *Project*: PersonalAIs: Generative AI Agent for Personalized Music Recommendations *Student*: Eleftherios Verouchis GitHub: TerryHive <https://github.com/TerryHive> | LinkedIn: lefteris-verouchis <https://linkedin.com/in/lefteris-verouchis> Email: verlef3 [ at ] gmail [ dot ] com | Location: Athens, Greece (GMT+3) First language: Greek | Languages: English (C2), German (B2), Norwegian (learning) ------------------------------ Personal Motivation As a Computer Science student with a passion for AI and user-centric applications, I find the PersonalAIs project particularly exciting. Music plays an emotional role in many people's lives, and combining that with conversational AI presents a creative and technically fulfilling challenge. This project is an excellent opportunity to integrate NLP, recommendation systems, and generative dialogue interfaces in a meaningful way. My background includes hands-on projects in machine learning, sentiment analysis, and backend APIs. I am also actively learning frontend development (React) to build complete interactive systems. Participating in GSoC through this project would help me sharpen my skills in full-stack AI applications while contributing to a tool that enhances everyday user experience. ------------------------------ Why GFOSS and this project? GFOSS’s commitment to open technology, accessibility, and innovation in real-world applications aligns with my values and interests. The PersonalAIs project is practical, emotionally resonant, and technically modern—bridging open data, AI, and user experience. Through this project, I hope to contribute to GFOSS’s mission while also working on something I truly enjoy: music and intelligent systems. I believe my technical background and enthusiasm will be a strong fit for the team and goals of this initiative. ------------------------------ Project Deliverables 🎧 What I’ll Build: - An AI-powered conversational agent that recommends personalized music playlists - Integration with Spotify API to fetch user data, liked songs, and preferences - Real-time playlist editing via chat (e.g., change genre, mood, tempo) - Option for users to use the system without linking personal data (exploratory mode) - UI with chatbot-like interaction (React or similar) 🛠️ Technologies: - NLP: Transformers (Hugging Face), BERT/GPT models - Backend: Node.js or Python (FastAPI), Express.js - Frontend: React.js (or Next.js), Chat UI components - APIs: Spotify Web API - Deployment: Docker, GitHub Actions (optional) ------------------------------ Timeline 🟨 Community Bonding (May 20 – June 16) - Study Spotify API and related user permission flows - Explore mood-based music tagging datasets (e.g. Moodify) - Prepare tech stack and architecture proposal - Discuss design with mentors ------------------------------ 🟩 Phase 1 (June 17 – July 15) - Week 1–2: Implement Spotify auth, fetch user profile & liked tracks - Week 3–4: Train/test mood & genre classification from conversation input - 📌 *Deliverable 1*: User login + basic chatbot interaction + static recommendations ------------------------------ 🟩 Phase 2 (July 16 – August 12) - Week 5–6: Integrate NLP with user input to refine playlist generation - Week 7: Enable live playlist editing ("make it more relaxing") - 📌 *Deliverable 2*: Personalized dynamic playlist creation based on chat inputs ------------------------------ 🟩 Phase 3 (August 13 – September 9) - Week 8–9: Build polished chatbot UI (React/Next.js) - Week 10: Finalize documentation, deploy demo, test edge cases - 📌 *Deliverable 3*: Functional conversational AI interface + full Spotify integration ------------------------------ Time Commitment I can dedicate ~30–35 hours per week to the project and will be fully available throughout the GSoC period without interruptions. ------------------------------ After GSoC I plan to maintain the project after GSoC, adding new features such as: - Voice input support - Alternative music sources (e.g. YouTube, SoundCloud) - Mood inference from facial expression/audio (stretch goal) - Contribution to the community through guides or tutorials ------------------------------ Experience & Fit 🎓 *Education*: - BSc in Informatics @ Athens University of Economics and Business - GPA: 7.35 - Exchange programs: NTNU (Norway) & Feng Chia University (Taiwan) 🛠️ *Projects*: - *IMDB Sentiment Classifier*: Python/ML model to classify reviews - *AccommoDate (Java)*: Scheduling tool with calendar logic - University-level experience in ML, Databases, Java/C++ 🔧 *Skills*: - Languages: Python, Java, C++, SQL, JavaScript - Tools: Git, Docker, Scikit-learn, HuggingFace, React (learning) ------------------------------ Convincing GFOSS My experience in machine learning, APIs, and user-facing tools makes me a strong candidate for this project. I’m eager to work with mentors to build something enjoyable and technically sound. I believe that combining AI with music can positively affect people’s lives and showcases the creative potential of open source AI. ------------------------------
Attachment:
Lefteris_Verouchis_CV.pdf
Description: Adobe PDF document
Attachment:
Lefteris_Verouchis_Transcript_of_Records.pdf
Description: Adobe PDF document
---- Λαμβάνετε αυτό το μήνυμα απο την λίστα: Λίστα αλληλογραφίας και συζητήσεων που απευθύνεται σε φοιτητές developers \& mentors έργων του Google Summer of Code - A discussion list for student developers and mentors of Google Summer of Code projects., https://lists.ellak.gr/gsoc-developers/listinfo.html Μπορείτε να απεγγραφείτε από τη λίστα στέλνοντας κενό μήνυμα ηλ. ταχυδρομείου στη διεύθυνση <gsoc-developers+unsubscribe [ at ] ellak [ dot ] gr>.