Lugano, Switzerland
Lorenzo Molinari
I'm a software engineer at adesso Schweiz, specializing in AI-assisted engineering, information security, cybersecurity and enterprise automation. My work bridges AI systems with security-first design, grounded in applied research on privacy-preserving machine learning. My master's thesis, recognized as one of Forbes Italia's Top Graduate 2025 theses, measured how explanation layers in federated learning systems leak private information and how differential privacy mitigates this at scale.
I write about the same throughline here: making AI systems that are both useful and actually private, not one traded off against the other. See about, projects or jump straight to writing.
About
I studied cybersecurity at the Università di Milano, then earned a double master's in Informatics between Università di Milano-Bicocca and USI Lugano. My research focuses on a central problem: how to build AI systems that explain their decisions to users without leaking private information in the process. My master's thesis, Privacy Leakage of Counterfactual Explanations in Federated Learning, demonstrates a practical approach to this challenge through federated learning with differential privacy. The writing section extends this work across broader topics in AI security and privacy engineering.
Today, I build AI-assisted engineering solutions and automation systems for adesso's clients: low-code development on multiple Roche projects, Power Automate BPA workflows, and security and privacy scans across other client codebases. I'm also part of adesso's AI Force, working to expand the company's AI capabilities. Previously, I split my time between leading R&D initiatives and working as a security engineer: designing cloud infrastructure, building RAG systems on Azure, developing security data pipelines with Microsoft Sentinel, Microsoft Defender and Microsoft Purview, and running RIOT phishing-awareness training. Earlier still, I managed event network infrastructure for the Bormio FIS Alpine Ski World Cup and built client websites at UnitBit, hands-on systems and front-end work that still shapes how I approach engineering today.
Career
-
Associate Software Engineer, adesso Schweiz AG, Lugano
Developer on multiple Roche projects, building Power Automate BPA and AI-assisted engineering workflows and running security and privacy scans across the codebase. Also part of adesso's AI Force, working to expand the company's AI capabilities.
-
System Engineer, R&D (Security & AI), 4IT Solutions, Gravesano
Cybersecurity and AI R&D: cloud proofs-of-concept, Microsoft Sentinel security data pipelines, RAG systems on Azure.
-
Teaching Assistant, "Ethics in Computing", USI Lugano
Student assistant under Prof. Silvia Santini, alongside the master's program.
-
Network Manager, Bormio FIS Alpine Ski World Cup, Bormio
Deployed and maintained high-availability event network infrastructure, including routers, firewalls, wireless access points, switches and broadcast systems, for international ski events, with zero-downtime connectivity for 10+ media vendors.
-
Web Developer, UnitBit, Bormio
Built and maintained responsive client websites, handling ongoing troubleshooting and performance improvements.
Education
-
Forbes Italia Top Graduate 2025
Named one of the 100 best Italian theses of the year for Privacy Leakage of Counterfactual Explanations in Federated Learning. Built a federated-learning system to measure how much private information explanation layers leak under membership-inference attacks, comparing standard FL against versions with counterfactual explanations and differential privacy (DP-SGD).
-
Double MSc Informatics, USI Lugano & Università di Milano-Bicocca
Joint double-degree program: one thesis, two conferrals. Spent the thesis year as a full-time exchange student at USI under UNIMIB's Student Mobility program. Thesis supervised by Prof. Marc Langheinrich (USI), with Dr. Dario Fenoglio as co-supervisor and Prof. Michele Ciavotta (UNIMIB) as the Bicocca-side supervisor.
-
USI Best Project Presentation, "Smart Angler"
Awarded for the Mobile & Wearable Computing course project: a fishing-session tracker with LLM-assisted fish-species recognition. Recognized for technical depth and practical integration of machine learning.
-
BSc Computer Systems & Networks Security, Università di Milano
Cybersecurity-focused bachelor's (Class L-31). Thesis: A Collaborative Editor for Text Annotation, supervised by Prof. Paolo Ceravolo: a real-time collaborative editor for flagging discriminatory text, built on Recogito and Firepad.
-
High School Diploma, ITIS Enea Mattei, Sondrio
Technological sector with Information Systems & Telecommunications specialization.
Projects
A curated selection out of a much larger pile of coursework and side projects. Source for all of them is on GitHub.
-
Master's thesis
Privacy Leakage of Counterfactual Explanations in Federated Learning
Measured how much a federated-learning explanation layer leaks about individual clients under membership-inference attacks, and what accuracy cost differential privacy pays to stop it. Forbes Italia Top Graduate 2025.
-
Bachelor's thesis
A Collaborative Editor for Text Annotation
Real-time collaborative editor for flagging discriminatory text, built on Recogito + Firepad.
-
Android · award winner
Smart Angler
Fishing-session tracker with LLM-assisted fish-species recognition. USI Best Project Presentation, Autumn 2024.
-
Solidity · Hardhat · React
DLoan
A decentralized loan-management app: Solidity smart contracts with an EthersJS/React frontend.
-
FastAPI · Mistral
Prompt Optimizer
Built for Mistral's "Moonshot Challenge": an Arena Mode that compares and judges prompt candidates across Mistral Small/Medium/Large.
-
Python · course project
K-Anonymity
Implementations of the Samarati and Mondrian k-anonymity algorithms, benchmarked against the UCI Adult and Bank Marketing datasets.
-
Java · distributed systems
Distributed Systems Assignment
Practical implementation of distributed computing concepts, including message passing, consensus algorithms and fault tolerance mechanisms for system reliability.
Writing
Long-form notes on privacy engineering and AI systems, the same throughline as the thesis, written for a general technical audience.
- How SOAR Systems Automate Security
Aug 10, 2026
A SIEM tells you your house is on fire; a SOAR system tries to put it out. Security Orchestration, Automation and Response (SOAR) platforms are the hands and feet of a modern Security Operations Center (SOC). They connect disparate security tools, automate tedious investigation tasks and stop cyber threats in real time without waiting for a human analyst.
- The Central System of Cybersecurity: SIEM Systems
Jul 27, 2026
Security teams usually drown in data. Every firewall, server and application generates millions of logs daily. A Security Information and Event Management (SIEM) system solves this. It is the central brain of an organization's infrastructure: it collects data from everywhere, normalizes it and uses analytics to find the actual cyber threats hidden in the noise.
- Fixing the Hallucination Problem: How RAG Systems Ground AI in Reality
Jul 13, 2026
Large Language Models predict text well but don't remember facts reliably. When they lack context, they "hallucinate" plausible, fake answers instead. Retrieval-Augmented Generation (RAG) fixes this by forcing the model to read specific, verified documents before it answers, instead of letting it guess.
- Universal Connectors for AI: The Model Context Protocol (MCP) and n8n
Jun 29, 2026
Early AI integrations required a custom connection for every combination of LLM and tool. Anthropic's Model Context Protocol (MCP) replaces that mess with a single open standard for how AI applications talk to external systems. Pair MCP with an automation platform like n8n and an agent gets access to thousands of tools without custom code for each one.
Contact
loremolinari1@gmail.com · WhatsApp · LinkedIn · GitHub · CV (PDF)