I’m an AI & Data Engineer at Acer EMEA, with 5+ years of combined experience across AI, data engineering, and analytics. My work spans enterprise agents, retrieval systems, MCP integrations, and Microsoft Fabric and Azure data engineering.
I enjoy taking an initiative from the first conversation to a working solution: understanding the task, choosing an architecture, building it, and making it easier to operate. That can mean an application, an automated workflow, or a better way to work with enterprise data.
My path started in civil engineering. At Politecnico di Milano, I graduated with 110 cum laude and applied neural networks and regression to urban crash data, leading to a peer-reviewed publication. Data consulting and enterprise BI led me to the AI work I do today.
How I work
01
Understand the task
Talk with business and IT teams to find the friction and define what a useful solution should do.
02
Design the solution
Choose the models, retrieval, tools, and data flows around the need and practical constraints.
03
Build and observe
Implement the workflow, include human review where needed, and make agent behaviour traceable.
Always learning
I follow developments in AI and test ideas in personal projects before deciding where they fit. My interests include agent architecture, AI observability, enterprise retrieval, and data platforms for AI.
Enterprise applications, agents, and data workflows I have designed, implemented, or helped deliver.
Enterprise projects
Production workflow
Product content review
An agentic workflow that checks e-commerce product content against regulatory documents and internal legal guidance, with legal and e-commerce teams reviewing the reports.
LangGraph · LangChain · Hybrid retrieval
Approach & contribution
The need
Help teams identify product claims that need attention and keep reviewing newly added products.
My contribution
Mapped the product database with its owner, designed the retrieval and agent workflow, and implemented analysis, review/scoring, and report-formatting stages.
The approach
Combined full-text and keyword retrieval with semantic vector search over reference documents. Grounded agents in those sources and added a second agent to review the analysis.
In use
The production workflow produces reports for human review. A reusable assistant skill also supports database-driven, on-demand runs.
Production RAG
Enterprise document retrieval
Summary-based retrieval that selects relevant documents first, then inspects the original sources when an answer needs more detail.
RAG · Embeddings · Document summaries
Approach & contribution
The need
Make a collection of large, varied enterprise documents easier to search and use in answers.
My contribution
Researched and proposed the architecture, secured manager agreement, and implemented it with the team.
The approach
Embedded document summaries linked to original sources. The workflow can answer from a summary or inspect the related source documents when the summary is insufficient.
In use
The custom retrieval approach has been operating for over a year.
Enterprise application
Marketing landing-page application
An Azure-hosted application that generates HTML landing pages from product information for the marketing team.
Azure · HTML generation · AI applications
Approach & contribution
The need
Give the marketing team another way to create product landing pages that had previously been handled manually.
My contribution
Architected and developed the application and its agentic workflow, using AI-assisted development, and hosted it on Azure.
The output
HTML landing pages for the team to work with.
Enterprise implementation
Agent tracing and observability
Tracing that makes agent execution, tool calls, throttling, and errors visible when investigating unexpected behaviour.
Application Insights · OpenTelemetry · Azure AI Projects SDK
Approach & contribution
The need
Debugging agent workflows required manually piecing together information from logs.
My contribution
Implemented agent tracing with Azure Application Insights and OpenTelemetry, applying experience with the Azure AI Projects SDK.
The result
The team can inspect what happens during agent calls and investigate tool usage, throttling, and errors with more context.
Enterprise integration
MCP tools for enterprise assistants
Approved third-party MCP servers and departmental tools connected to an enterprise assistant through its existing user configuration framework.
MCP · Docker · REST APIs · Azure Functions
Approach & contribution
The need
Make hosted and remote tools available to the teams that need them within the assistant.
My contribution
Containerised and hosted third-party MCP servers, implemented agent clients, and integrated departmental tools exposed through Azure Functions and REST APIs.
The boundary
Worked within the team's existing control-tower and Redis configuration framework. Tool approval and access configuration were already managed by the team.
Reusable IT workflows
Fabric model and data workflows
MCP-assisted workflows for semantic-model maintenance and targeted investigation of data discrepancies in Microsoft Fabric.
Microsoft Fabric · Power BI Modeling MCP · Notebooks
Approach & contribution
The need
Support IT colleagues maintaining semantic models and investigating pipeline issues without opening these operations to general business users.
My contribution
Designed reusable workflows using Fabric and Power BI Modeling MCP tools for model changes and data investigation.
An example
Investigated missing ticket records with a supplementary notebook, retrieving the missing data while keeping the original pipeline intact.
Personal projects & labs
Experiments and learning outside my enterprise work.
Personal project
Budgeting assistant
Designed a personal budgeting assistant and tried open-source models from Hugging Face alongside frontier models to explore which models fit a straightforward, defined task.
Hugging Face · Model comparison
Learning lab
Azure document retrieval
Tested extraction of personal documents containing layouts and tables, then integrated the extracted content with Azure AI Search. A personal lab exploring document retrieval.
Document extraction · Azure AI Search
Personal knowledge system
Obsidian second brain
A connected space for learning notes, ideas, books, and project tracking. I have also designed a personal assistant to help organise tasks and ideas in this system.
Obsidian · Knowledge management
Personal website
This portfolio
A lightweight portfolio built with HTML, CSS, and JavaScript, with an emphasis on clear writing, accessible navigation, and readable project descriptions.
AI delivery today, built on data engineering and business intelligence.
Mar 2025 – Present
AI & Data Engineer
Acer EMEA · Switzerland
Interview business and IT stakeholders to identify workflow needs and design AI applications and automation solutions.
Built a production LangGraph workflow to review e-commerce product content against regulatory documents and internal legal guidance, combining keyword/vector retrieval, agent review and scoring, and reports reviewed by legal and e-commerce teams. Enabled database-driven, on-demand execution through an assistant skill.
Architected and developed an Azure-hosted application generating HTML landing pages for marketing, using AI-assisted development.
Researched, proposed, and implemented with the team an enterprise RAG architecture using embedded document summaries linked to original sources, with deeper document inspection when needed.
Implemented agent tracing with Azure Application Insights and OpenTelemetry, providing visibility into agent execution, tool calls, throttling, and errors for troubleshooting.
Containerised and hosted third-party MCP servers, implemented agent clients, and connected departmental tools to the assistant's existing configuration framework.
Designed Fabric and Synapse ETL pipelines and transformation notebooks, and implemented deployment pipelines across development, testing, and production.
Designed and maintained semantic models, and built reusable IT workflows using Fabric and Power BI Modeling MCP tools for model maintenance and targeted data recovery, including supplementary notebooks to retrieve missing ticket records.
Sep 2023 – Mar 2025
Senior Business Intelligence Specialist
Gi Group Holding · Milan, Italy
Delivered Power BI dashboards and semantic models supporting financial, operational, and market reporting.
Developed data ingestion and transformation workflows using Python, SQL, Azure Synapse, and Microsoft Fabric, including pipelines and notebooks.
Collaborated with international business stakeholders to define reporting requirements, automate KPIs, and deliver BI initiatives.
Jul 2021 – Aug 2023
Consultant - Data Analyst
CitiEU | CitiME · Milan, Italy
Developed Python ETL workflows integrating TomTom MOVE data into SQL databases for transportation analytics.
Built API-driven maps providing traffic insights and accident notifications, and trained client teams in using data tools.
Jan 2018 – Jul 2018
Assistant Project Manager (Internship)
Faraz Rahe Hasti · Tabriz, Iran
Infrastructure coordination, risk analysis, and site oversight for engineering projects.
Toolkit
Skills
AI engineering and enterprise data, with the tools to connect them.
AI & agentic systems
Solution architecture, agent workflows, and applications built around business needs.