Open to opportunities

Sofia Casarin

@sofiacasarin

Message

I build production-ready multimodal AI systems and efficient computer vision models.

Italy

Message

What I'm looking for

I'm particularly interested in computer vision, vision-language models, efficient foundation models, and production-ready generative AI systems built with MLOps and cloud-native infrastructure.

At the Division of Language and Vision Machine Learning in Rome, I build enterprise-scale generative AI systems for intelligent document understanding and knowledge management. I design production-ready multi-agent systems, retrieval-augmented generation pipelines, multimodal ingestion workflows, and cloud-native AI services on AWS.

My research spans computer vision, neural architecture search, video understanding, medical imaging, and efficient deep learning. I hold a Ph.D. summa cum laude in Computer Science from the Free University of Bozen-Bolzano and have published work at CVPR, ICLR, ICCV workshops, and ICML workshops.

I've also worked with the University of Padua and Sony Europe, developing robust architecture optimization methods under distribution shifts. My work includes winning first place in the CVPR NAS Challenge 2023 across 10 of 11 datasets.

Experience

Work history, roles, and key accomplishments

Current

Applied AI Engineer

Current

Applied AI Engineer

Mar 2025 - Present (1 year 6 months)

Designed novel zero-cost metrics and predictors for division of labor in machine learning, based on NAS algorithms for robust architecture optimization under distribution shifts. Worked within the AI division on enterprise-scale generative AI systems for intelligent document understanding.

Computer Vision Deep Learning Multimodal AI PyTorch AWS Docker RAG MLOps

Education

Degrees, certifications, and relevant coursework

Free University of Bozen-Bolzano

Doctor of Philosophy, Computer Science

2022 - 2025

Grade: Summa Cum Laude

Ph.D. in Computer Science, graduated summa cum laude. Research focused on neural architecture search, computer vision, and efficient deep learning.

University of Padua

Master of Science, Telecommunications Engineering

2019 - 2021

Grade: Summa Cum Laude

M.Sc. in Telecommunications Engineering, graduated summa cum laude. Thesis on predictor-based neural architecture search under data distribution shift.