Marco Santoni
Data Engineering Manager
Milan, Italy
Actions
Marco Santoni is a Data Platform and Engineering Manager at TeamSystem, where he works on scalable data platforms, AI-enabled data products, and data governance architectures. Based in Milan, he is an active member of Python Milano community and co-host of the Intervista Pythonista podcast. He is also involved in education as Technical and Scientific Advisor at ITS Rizzoli, where he teaches data and software engineering topics.
Area of Expertise
Topics
Why is our project late?
Why is it so hard to respect deadlines for a dev team? Is it just a wrong planning methodology? Are there fundamental limits in the way our brain makes plans? Has it always been like this?
We’ll investigate the way humans perceive plans and unexpected events. Do we have biases in estimates?
Chatting with Data, Safely: Role-Based Authorization for Text-to-SQL Agents
Text-to-SQL agents can query anything — but should they?
When AI agents turn natural language into executable SQL, traditional data-access guardrails are easy to bypass. The challenge becomes critical when real users, with real roles and permissions, are the ones chatting with your data.
In this talk, we explore how to decouple AI agent logic from authorization policies without sacrificing flexibility or performance. Starting from the real-world challenges of Text-to-SQL in agentic platforms, we ask: how can we ensure users see exactly the same data—no more, no less—whether they access it through a UI or via an AI agent?
We’ll present a production-ready solution combining:
- a Python package that inspects and rewrites the SQL Abstract Syntax Tree (AST), and
- partial evaluation results from Open Policy Agent (OPA) to enforce fine-grained authorization at query time.
This approach keeps authorization logic centralized, auditable, and independent from LLM prompts, while integrating seamlessly with Text-to-SQL tools.
You’ll leave with a concrete architectural pattern, open-source building blocks, and practical lessons on how to safely scale AI agents—without turning your LLM into a superuser.
Scaling Data Mesh Orchestration with Dagster: Platform-Driven DAGs without Platform Friction
Data Mesh promises scalability and autonomy, but orchestration often becomes the hidden bottleneck: as data products grow, cross-domain dependencies explode, DAGs become fragile, and upstream teams are forced to react to every new consumer.
In this talk, we present a platform-designed orchestration solution built on top of Dagster that enables company-wide orchestration without coupling data products together.
Instead of requiring data engineers to manually define DAGs or modify upstream pipelines, each data product simply declares its dependencies on upstream output ports. From these declarations, the platform automatically builds and maintains the global DAG behind the scenes, ensuring:
- upstream data products remain completely unaffected by new consumers
- orchestration scales as the number of data products grows
- platform standards are enforced without slowing down teams
We’ll walk through:
- the orchestration challenges we faced when adopting data mesh at scale
- how Dagster was extended and shaped into a platform abstraction, not exposed directly to product teams
- the Python-based design patterns used to dynamically generate and evolve DAGs
- the resulting developer experience for data engineers building data products
This session is a practical, production-tested story of how Python and Dagster can be used to enable scalable orchestration in a real data mesh implementation — without turning the platform team into a bottleneck.
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top