Felice Schena

Deep Learning Engineer

Modena, Italy | feliceschena@proton.me | feliceschena.com | github.com/FeliceSchena | linkedin.com/in/felice-schena

Summary

I'm a Deep Learning Engineer working on vision-language models and generative AI inference. I wrote my master's thesis on fine-tuning CLIP and SigLIP for domain-specific image retrieval, and that work carried straight into production at Tetra Pak, where I now own the GenAI and ML inference architecture for Tetra Pak Factory OS. Along the way I picked up real Kubernetes and MLOps ability, so the models I build are also the ones I ship.

Experience

MLOps Engineer

Jul 2025 – present

Tetra Pak — Modena, Italy

Owning the Generative AI and machine learning inference architecture for Tetra Pak Factory OS, on-prem and at cloud level
Serving models with vLLM and building the CD pipelines with FluxCD and Kubernetes that get them into production

AI & Analytics Engineer

Feb 2025 – Jul 2025

FEAB Sistemi (Tetra Pak Consultant) — Modena, Italy

Building machine learning models for Asset health monitoring, moving them towards production use in quality control.
Designing and maintaining analytics for the Development & Technology department.
Designing RAG Flows for internal Documents.

Artificial Intelligence Intern

Aug 2024 – Jan 2025

Tetra Pak — Modena, Italy

Researched and implemented state-of-the-art vision-language models (CLIP, SigLIP) and adapted them to Tetra Pak's quality-control use case. Patent
Manually annotated a dataset of ~1,000 image–text pairs and fine-tuned models with transfer-learning techniques.

Projects

Ovarian Cancer Survival Prediction — AI for Bioinformatics project, UNIMORE (Feb 2024). Graph Neural Networks with attention layers integrating genomic and clinical data to improve survival-prediction precision for cancer patients.

LocoBot — Computer Vision & Cognitive Systems exam, UNIMORE (Jul 2023). Fine-tuned YOLOv5 on the HaGRID dataset to recognize hand gestures for robot tracking and stopping, without losing the network's original knowledge.

Car Lo — IoT university project, UNIMORE (Feb 2023). IoT system promoting eco-friendly driving: Arduino + IMU + GPS hardware, an MLP classifying driving style in real time, and a Flask app on Google Cloud for live feedback.

MMR Driverless — Formula Student — MoRe Modena Racing (Oct 2021). Team leader for the autonomous-driving division of a Formula Student team: sensor system design, Stanley controller for trajectory tracking, skidpad optimization.

Education

MSc, Computer Science Artificial Intelligence

2022 – Apr 2025

University of Modena and Reggio Emilia — Modena, Italy

Thesis: Fine-tuning Vision-Language Models for Domain-Specific Image Retrieval with Small-scale Datasets.

BSc, Computer Science

2017 – 2021

University of Modena and Reggio Emilia — Modena, Italy

Thesis: Variable Neighborhood Search — optimal placement of microservices in fog architectures.

Certifications & Courses

AWS Solution Architect

Oct 2021

Tree srl

Course on designing scalable, highly available and fault-tolerant systems on AWS, with a focus on cost constraints and Quality of Service.

Skills

ML / Deep Learning: PyTorch, Hugging Face(accelerate-transformers-peft), vLLM, Pandas, NumPy, Scikit-Learn, Plotly, Streamlit
AI: VLMs, LLMs, RAG, MoE, Fine-tuning, Agents
Coding: Python, C++, C, Java
MLOps / Infra: Docker, Kubernetes, FluxCD, CI/CD pipelines, Azure, AWS
Databases: PostgreSQL, MySQL, SQLite, Weaviate, PGVector
Languages: Italian (native), English

Publications

Placement of IoT Microservices in Fog Computing Systems: A Comparison of Heuristics — Algorithms (MDPI), open access, 2021. mdpi.com/2478364