Matteo Massetti
Senior Machine Learning Engineer
Senior Machine Learning Engineer with 5+ years of experience, specialized in agentic AI and multi-agent systems. Strong engineering background across full-stack development and cloud platforms, applied to building production-grade solutions for enterprise clients in regulated sectors. Experienced in technical leadership, client-facing delivery, and mentoring engineers.
Contact Me Download CVWork Experience
Senior Machine Learning Engineer
Machine Learning Reply, Milan | Jan. 2026 - Present
Leading the design and delivery of end-to-end ML and GenAI solutions for enterprise clients on Google Cloud and AWS, defining the technical architecture and ensuring quality and scalability from PoC to production. Technical reference point for a business unit of about 20 people; managing and mentoring 4-5 junior engineers and contributing to pre-sales and technical proposals.
Key projects
- Architected the core multi-agent logic of an agentic platform for a major Italian media and telecommunications group, with RAG over tens of thousands of heterogeneous documents from the corporate intranet (SharePoint and other sources).
- Designed the real-time interaction pipeline of a digital human brand ambassador built on the Gemini Live API: low-latency bidirectional audio, barge-in handling and avatar integration, keeping responses consistent with the brand's tone and knowledge.
- Leading quality tuning of a LiveKit and Mistral voice assistant for a Greek financial services multinational, improving the audio layer and agent logic to reach production-grade quality.
- Built a GenAI-based ingestion pipeline on AWS that extracts structured data from emails and attachments for downstream workflows.
Machine Learning Engineer
Machine Learning Reply, Milan | Jan. 2024 - Dec. 2025
Designed and delivered ML and GenAI solutions for enterprise clients in different sectors, mainly on Google Cloud.
Key projects
- Built a multi-agent Text-to-SQL system on Google Cloud that lets a procurement department in the media industry query a relational supplier database in natural language, handling multi-table schemas and generating charts to surface insights from the results.
- Developed a GenAI assistant for guarantees analysis for a client in the travel and tourism industry, processing multiple data sources to identify insurance and contractual coverage across international regulations.
- Developed near-real-time anomaly detection on the data stream feeding the OTT platform of a media client, monitoring the health status of video and digital channels and flagging anomalies.
- Designed and delivered 10 workshops on cutting-edge AI technologies (multi-agent architectures, multimodal models, reinforcement learning) for a major Italian bank, tailored to specific business use cases, with 20-25 participants each.
Software Developer
Micra s.r.l., Corciano | Jul. 2021 - Jan. 2024
Full-stack developer of products, services and platforms in the banking and industrial sectors, using Python and Java. Within the Research and Development group, I contributed to the study of new technologies, including Machine Learning and Artificial Intelligence, aimed at improving the company's products and identifying new possible solutions for customers.
Key projects
- Trained an ML model to read handwritten forms for a major luxury fashion company, automating the digitization of paper-based data.
Stage
Università degli Studi di Perugia, Perugia | 2018
Programming of a plug-in for the Moodle platform, which could facilitate and automate the administration and correction of multiple-choice questionnaires in order to test the knowledge of a large group of university students efficiently.
Personal Projects
MoneyTrack
An Android application that allows the user to track and review all their earnings and expenses. The main goal of MoneyTrack is to be simple to use, to provide a quick way to enter a new expense or income (in app called Bids) and an intuitive category management.
Link →Leveraging Physical Cues for Learned Representations in Visual Question Answering
Inferring knowledge from various sources and data, such as natural language and visual data, is challenging. This work presents a new module for neural network architectures aimed at improving this capability.
Link →Brain Tumor Detection
In this project, I implemented a convolutional neural network to perform semantic segmentation of MRIs depicting brains.
Link →Irony and Sarcasm detection in tweets
In this project, developed in collaboration with a colleague from the University of Pisa, I implemented several machine learning models to solve the tasks of the EVALITA challenge.
Link →Low Rank Approximation with Normal Equation and Cholesky Factorization
In this project, developed in collaboration with a colleague from the University of Pisa, we used an altered optimisation algorithm to calculate a low rank approximation of a matrix A.
Genetic Algorithm and Parallelism for the Travelling Salesman Problem
In this project, I implemented a genetic algorithm to solve the travelling salesman problem on a parameterisable number of nodes. To speed up the evolution process, I used and compared two parallelisation techniques: C++ threads and the FastFlow library.
Link →Education
Master's Degree in Computer Science
Università di Pisa, Pisa | Sep. 2019 - Jul. 2023
Artificial Intelligence curriculum - Final mark: 110/110 cum laude
Completed while working full-time
Thesis: Leveraging Physical Cues for Learned Representations in Visual Question Answering
Bachelor's Degree in Computer Science
Università degli Studi di Perugia, Perugia | Sep. 2016 - Sep. 2019
Final mark: 110/110 cum laude
Thesis: Multidimensional Link Prediction
Diploma in Administration, Finance and Marketing
Istituto R. Bonghi, Assisi | Sep. 2011 - Jun. 2016
Final mark: 98/100
Certifications
Miscellaneous
Articles written for QUID, Mensa Italia's digital magazine. Each volume is monographic and addresses a topic in a multidisciplinary manner, from diverse and often surprising perspectives.
L’infinito che ci rassicura e assopisce
QUID 16: La fine
How endless entertainment lets us escape awareness of our own finitude, drawing on Lem's Solaris.
Leggi l'articolo →Questione di Scale
QUID 14: Intelligenza Artificiale
Argues that the most imminent risk of AI is humans using it to amplify existing harms, such as political micro-targeting, fraud and disinformation, through sheer scale.
Leggi l'articolo →Skills
GenAI & Agents
LangGraph, LangChain, Google ADK, MCP, A2A, RAG, Prompt Engineering, Fine-tuning, LLM Evaluation
Models & APIs
Gemini (incl. Live API), Claude, GPT, Mistral, Hugging Face, LiveKit
Machine Learning
PyTorch, scikit-learn, Computer Vision, OCR, Anomaly Detection, Time Series
Cloud
Google Cloud (Vertex AI, BigQuery, Cloud Run), AWS (Bedrock, Lambda, Step Functions), Azure (Foundry)
Engineering
Python, SQL, HTML, JavaScript, CSS, Java, FastAPI, Docker, CI/CD
Tools
Git, Claude Code, Jupyter