I adapt vision-language models and machine-learning systems to domain-specific data. And deploy them in Kubernetes clusters.

experience

• type work date 2025-07 → present location Modena, Italy

  MLOps Engineer

  Tetra Pak

  Full-time MLOps engineer,working for the Tetra Pak Factory OS.

  FluxCDPyTorchvLLMKubernetesCKADAnalytics

  read more

  Since July 2025 i’ve started to work as DevOps Engineer in Tetra Pak. Due to my strong background in Artificial Intelligence, the work took a MLOps path. Implementing the new inference architecture for Generative AI Inference, on-prem and/or at Cloud Level.

• type work date 2025-02 → 2025-07 location Modena, Italy

  AI & Analytics Engineer

  Tetra Pak

  Full-time AI & analytics engineer, external consultant at Tetra Pak (employed by Feab sistemi srl).

  VLMPyTorchAnalytics

  read more

  Since February 2025 I work full time at Tetra Pak as an AI & Analytics engineer, as an external consultant employed by Feab sistemi srl.

  The role continues the vision-language model work I started as an intern — adapting models such as CLIP and SigLIP to domain-specific industrial data — and extends it with analytics engineering for the Development & Technology department.

• type work date 2024-08 → 2025-01 location Modena, Italy

  Artificial Intelligence Intern

  Tetra Pak

  Six-month AI engineering internship in the D&T department: adapting vision-language models to domain-specific data for automated quality control.

  CLIPSigLIPFine-tuningPyTorch

  read more

  Tetra Pak is a Swedish multinational specializing in food packaging and processing solutions, operating in over 160 countries with more than 25,000 employees.

  During my internship as an AI engineer in the D&T department, I focused on adapting vision-language models to domain-specific data, to improve the efficiency of the company’s quality-control process by automating the test environment. I researched and implemented state-of-the-art models such as CLIP and SigLIP, and devised ways to adapt them to Tetra Pak’s specific use case.

  The work ran the full length of a real ML project: data preprocessing, manual annotation of a dataset of about a thousand image–text pairs (a lesson in how much data quality matters), and fine-tuning with transfer-learning techniques to improve performance on domain-specific data.

  At the end of the internship I presented my findings and recommendations to the team, highlighting the potential benefits of integrating vision-language models into the quality-control process. The work was well received and led to my current full-time role.

Full career — education, courses, skills →

selected projects

All projects →

writing

• 2026-07-04 Sample post — formatting reference (keep as Draft)

  Not a real post — shows which Notion blocks the sync understands.