Turin, Italy · Portfolio 2026
Machine Learning Engineer
GenAI & ML Engineering · AI Delivery Center, Intesa Sanpaolo
I build data pipelines, fine-tune LLMs and orchestrate agents, from prototype to production. Most of the work sits between the models and the plumbing around them: retrieval quality, evaluation and inference that has to run outside a datacenter.
01 Work
Applied research with an evaluation attached. Five projects, one case study in detail; the rest point to the code.
Featured project
A LangGraph multi-agent system that aligns business documents with relational schemas in a Neo4j knowledge graph, then answers questions over that alignment with hybrid retrieval and a grading step that refuses ungrounded answers.
Why this one Selected because it is the most fully documented piece of work here: two coordinated graphs, an evaluation protocol and reported results.
Builder graph
Query graph
0 hallucinations · 4.31/5 mean score · 21 ablation studies
Glossaries, technical documents and relational schemas describe the same data in different vocabularies. The thesis asks whether a multi-agent system can hold that alignment in a knowledge graph and answer over it without inventing facts.
Thesis work as a graduate intern at Data Reply: the builder and query graphs, the retrieval stack and the evaluation protocol. Evaluated on seven datasets, with ablations for the retrieval and validation stages.
02 Experience
Industry, academic honours and applied research, in reverse order.
Sep 2026 — Present
Intesa Sanpaolo · AI Delivery Center · GenAI & ML Engineering team · Full-time
Design, develop, ship and maintain Machine Learning, Generative AI and Agentic AI solutions end-to-end, from architecture and data pipelines to production deployment and optimization. Champion AI-assisted coding and AgenticOps across teams, building cross-cutting AI services with FastAPI on Databricks.
Feb — Aug 2026
Data Reply · Turin, Italy · Hybrid
Designed production-oriented Generative AI for metadata management and semantic discovery. Architected scalable AI workflows combining LLM reasoning and backend services, from prototype toward deployment.
Oct 2024 — Jun 2025
IEEE-Eta Kappa Nu · Mu Nu Chapter · Turin, Italy · Volunteer
Recruitment and member engagement for the IEEE-HKN electrical and computer engineering honour society. Organized technical events and workshops.
Feb — Jun 2024
University of Enna “Kore” · Enna, Italy · On-site
Built ML pipelines for dynamic gesture recognition on Leap Motion time-series data. Statistical feature analysis; contributed to the DYLEM-GRID dataset.
2024 — 2026
Politecnico di Torino
110/110Thesis: “Generative AI and Foundation Models for Automated Data Engineering and Metadata Orchestration”.
2021 — 2024
University of Enna “Kore”
110/110 cum laudeThesis: “Recurrent Neural Networks for Dynamic Gesture Recognition”.
03 Profile
I work where LLM systems meet the unglamorous parts.
I’m Marco. I’m from Agrigento and based in Turin. My work sits between models and the plumbing around them: retrieval, reranking, evaluation and quantization for hardware that isn’t a datacenter.
I care about systems that can be checked — grounded answers, measured retrieval, honest ablations — more than demos that only work once. That preference runs through the thesis, the SemEval system and the way I evaluate the AI services I ship at work.
04 Contact
An opportunity, a collaboration, or a question about something on this page.
marcantonio.lopez@ieee.orgEmail is the fastest route — I usually reply within a day.