Experience

  1. AI Research Scientist Intern

    Meta FAIR

    Core member of the Omnilingual Team for 15 months.

    Led the omnilingual extension of SONAR to 4,000+ language varieties via a two-stage teacher-student distillation framework, achieving 15x cross-lingual error reduction across 1,560 languages. arXiv paper

    Also contributed to Omnilingual MT (first MT system for 1,600+ languages), BLASER 3 (most multilingual MT quality estimation model to date), and multilingual data curation pipelines at the scale of billions of examples. arXiv paper

    Earlier, proposed a character-level encoder via teacher-student distillation achieving state-of-the-art zero-shot speech translation on FLEURS, surpassing SeamlessM4T. ACL 2025 paper

  2. AI Research Scientist Intern

    Apple
    Contributed to the Machine Translation team within the AI/ML division, primarly focused on Prosody Evaluation and Speech Translation. My research, which was published in WMT 2024, showed that although current state-of-the-art Speech Translation systems do have a understanding of prosody, they rarely leverge it in practice, indicating significant room for improvement. WMT 2024 paper

  3. AI Research Scientist Intern

    Dolby

    Proposed a sequential contrastive framework for audio-visual learning using multidimensional sequential distances over non-aggregated representations, achieving up to 3.5x relative improvement in retrieval recall over previous methods. ICASSP 2025 paper

    Contributed to a masked generative video-to-audio model combining a full-band audio codec with sequence-to-sequence parallel generation, achieving both high audio quality and temporal synchronicity with visual actions. ECCV 2024 paper

  4. AI Research Scientist Intern

    Zeta Alpha
    Did research on ranking and summarization, and built a transformer-based pairwise passage reranker and a generative QA model deployed in the first version of Zeta Alpha’s semantic search engine.

Education

  1. PhD in Artificial Intelligence

    UPC Barcelona
    Supervised by José A. R. Fonollosa and Marta R. Costa-jussà, I researched several topics within Speech Translation, including speech segmentation, data augmentation, efficiency inference, transfer learning, and zero-shot learning, and published in top-tier conferences including ACL, EMNLP, ICASSP, and InterSpeech. Thesis: Robust and Data-Efficient End-to-End Speech Translation

  2. MSc in Artificial Intelligence

    University of Amsterdam
    Graduated cum laude with a GPA of 8.3/10. Through advanced elective courses, I specialized in sequence modeling, deep learning, transformers, and NLP. Thesis: Complex Question Answering by Pairwise Passage Ranking and Answer Style Transfer. → Thesis PDF

  3. MSc in Quantitative Finance

    VU University Amsterdam
    GPA: 7.7/10. My thesis titled “An ensemble of Recurrent Neural Networks for high-frequency stock market classification” lead to a publication in the Journal of Forecasting. → Thesis PDF

  4. Diploma in Accounting and Finance

    Athens University of Economics and Business
    GPA: 7.1/10 with a specialization in Quantitative Analysis and Financial Modeling.
Skills
Infrastructure
Distributed Training

FSDP, ZeRO, mixed-precision training, data pipelines at scale, multi-node GPU clusters, Slurm
Frameworks

Python, PyTorch, torchaudio, HuggingFace, fairseq1, fairseq2, Weights & Biases, polars
Research Areas
Core

Multilingual & Multimodal Representation Learning, Large Language Models, Contrastive Learning, Self-supervised Learning, Teacher-Student Distillation, Knowledge Distillation
Methods

Cross-modal Alignment, Sentence Embeddings, Optimal Transport, Zero-shot Transfer, Dense Retrieval, RLHF, DPO, Mixture of Experts, Parameter-Efficient Fine-tuning
Applications

Speech Translation, Machine Translation, ASR, Audio-Visual Learning, Automatic Quality Estimation
Teaching Experience

Artificial Intelligence with Deep Learning (M.Sc.)

UPC Barcelona — 2022, 2023, 2024

Coordinated the lectures and the practical labs for the sequence-to-sequence modeling and NLP parts of the course. → Course Page

Machine Learning for the Quantified-Self (M.Sc.)

VU University Amsterdam — 2018, 2019, 2020

Supervision of practical labs, assisting students with python and ML questions, grading assignments and final exams. → Course Page

Introduction to Data Science (B.Sc.)

VU University Amsterdam — 2019, 2020

Supervision of practical sessions, grading assignments and final exams. → Course Page

Academic Supervision

Investigating Dropout in Encoder-Decoder Transformers

UPC Barcelona — 2023

BSc Thesis Supervisor for Pol Rosinés Pozo → Thesis PDF

Context-Aware Speech Translation

UPC Barcelona — 2022

BSc Thesis Supervisor for Miquel M. de Morentin Cardoner → Thesis PDF

Music Generation with Deep Learning

UPC Barcelona — 2022

Research Project Supervisor for Pau Lozano García

Academic Service
  • Program Committee Member: International Conference on Spoken Language Translation (IWSLT)
  • Reviewer: ACL Rolling Review, ICASSP, Interspeech, and various ACL-affiliated venues

Projects

AI DJ Project | 2021

This project features an AI DJ system designed to perform a collaborative back-to-back set with a human artist. Uniquely, the system is sound-agnostic; it uses a language model to understand the sentiment of YouTube comments to find songs with a similar "feeling" to the recent track history. The project was showcased at the various events including S+T+ARTS Festival in Barcelona and Sonar+D in Lisbon.

Awards
Cum Laude Honors
University of Amsterdam ∙ August 2020
Awarded for high academic achievement in the M.Sc. in Artificial Intelligence program.
Languages
100%
English
100%
Greek
25%
Spanish
25%
German