CV

AI/ML Engineer Consultant

Live Reply

Generative AI consulting focused on enterprise multi-agent systems, knowledge management, agentic analytics, and cloud infrastructure.

• Delivered an enterprise GenAI knowledge-management system over 200+ technical documents, reducing operator search time by 52% through ingestion pipelines, retrieval evaluation, Qdrant, Neo4j, and agentic GraphRAG workflows.
• Designed an Azure-based agentic analytics platform that converts transactional data into validated, actionable reports, reducing recurring report preparation time by 81% through SQL and object-storage integration, specialized agents, workflow memory, and automated quality checks.
• Built a stateful multi-agent orchestration platform with Python and LangGraph, enabling non-technical users to execute AI-assisted business processes and reducing end-to-end execution time by approximately 50%.
• Implemented reusable agent integrations using tool calling, MCP-compatible interfaces, structured Pydantic schemas, and human-in-the-loop workflow stages across 4 engineering teams.
• Provisioned reproducible AWS environments with Terraform, Kubernetes, networking, load balancing, API gateways, and containerized AI services, supporting separate development, testing, and production configurations.
• Led GenAI coding-platform adoption across 4 engineering teams by defining agent workflows, repository-aware development practices, and reusable tool integrations for multiple software-delivery use cases.

Applied AI Researcher, Short-Term Collaboration

BeyondShape

• Delivered 2 applied-AI pipelines: a 3D spine-reconstruction workflow using gradient boosting, convolutional networks, and generative methods, and an isolation-forest anomaly detector for large-scale ISIC dermatology data.

Applied AI Researcher

University of Tübingen, Mental Health Mapping Lab

Applied AI research in psychiatric neuroimaging, generative modeling, explainable AI, and medical-image analysis.

• Built a reproducible PyTorch and SLURM pipeline for distributed medical-image training, inference, experiment tracking, and explainability evaluation across approximately 70,000 DICOM/NIfTI images and 300,000 clinical records.
• Developed generative-AI and explainable-AI methods for psychiatric neuroimaging, combining diffusion-based counterfactual generation, computer vision, statistical evaluation, and clinically oriented model interpretation.
• Contributed to 2 open-source medical-AI projects covering foundation-model merging and neuroimaging explainability, including MedSAMix and HDPS.

DevOps Engineer, Telecommunications Consultant

Hewlett Packard Enterprise

DevOps and consulting role delivering CI/CD, backend integration, and delivery automation for a major telecommunications customer.

• Built a Jenkins, Java, Spring, and SQL deployment system for 1 major telecommunications customer, reducing release execution time by approximately 90% while improving repeatability and operational reliability.
• Owned delivery across 3 stakeholder groups---client representatives, software developers, and operations engineers---translating business requirements into technical designs, implementation plans, and production releases.
• Implemented Java and SQL backend integrations and introduced containerized delivery components, removing production blockers across multiple release cycles.

IT Consultant

Concept Solution

• Supported 4+ digital-transformation initiatives spanning robotic process automation, customer profiling, blockchain traceability, and e-commerce recommendation systems by translating business requirements into technical solution designs.

Bachelor of Science

University of Naples Federico II

Computer Science and Engineering

• Computer science, mathematics, and physics background, including databases, software engineering, and cybersecurity

Master of Science

Politecnico di Milano

Computer Science and Engineering

• Artificial Intelligence focused path
• Thesis on Reinforcement Learning

MedSAMix: A Training-Free Model Merging Approach for Medical Image Segmentation

AAAI / arXiv

Authors: Yang Y.*, Su G.*, Hu J.*, Sammarco F., Geiping J., Wolfers T.

Lightweight Model for Session-Based Recommender Systems with Seasonality Information in the Fashion Domain

ACM RecSys Challenge

Authors: Della Volpe N.*, Mainetti L.*, Martignetti A.*, Menta A.*, Pala R.*, Polvanesi G.*, Sammarco F.*, et al.

ACM RecSys Challenge and Publication

AI Summer School Instructor