PhD Researcher · Computer Vision · Semantic Future Prediction · World Models · Generative Modeling
Contact Email: e.karypidis@athenarc.gr
I build systems that predict the future — specifically, how visual scenes will evolve over time. My PhD focuses on semantic future prediction and latent-space world models, using transformers, self-supervised learning, and vision foundation models. I am a PhD Candidate at NTUA and fellow of Archimedes Research Unit / Athena RC.
- Semantic future prediction & world models — forecasting world evolution in learned latent spaces
- Vision foundation models — leveraging and extending large-scale pretrained visual representations
- Generative models × self-supervised learning — bridging generation and representation learning
- Machine unlearning — selective knowledge removal in deep networks