Savvas Tsimperis

At NTUA’s Artificial Intelligence and Learning Systems Lab, I’m building a Python framework to study how pre-play communication shapes cooperation and collusion among LLM agents. I’m using game-theoretic simulations, including price competition, the Iterated Prisoner’s Dilemma, and sealed-bid auctions.

For a multimodal movie recommendation project at NTUA, I built pipelines for poster and plot-summary classification and user–movie link prediction. Fine-tuning ResNet-18 improved poster-classification Macro F1 from 0.14 to 0.26, while a BiGRU with GloVe embeddings and attention raised plot-summary Macro F1 from 0.29 to 0.41.

I also developed a Variational Autoencoder pipeline for melanoma anomaly detection, reaching 0.79 ROC-AUC with a patient-safe data split. In a separate big-data project, I implemented and benchmarked Apache Spark queries over HDFS on Kubernetes.

My background includes an M.Eng. in Electrical and Computer Engineering from the University of Thessaly, and I’m currently pursuing an M.Sc. in Data Science and Machine Learning at the National Technical University of Athens. I’m seeking a Machine Learning / AI Engineer role where I can apply rigorous experimentation and modeling to real-world AI products.