Before reading the description of the event and RSVPing, please read the following message:
Please fill out this form to attend the AI Native Meetup on July 2nd. The information you provide will be used strictly for Cisco’s security requirements and will be deleted after the event. Once you complete the form, your name will be added to the official attendee list and you’ll be allowed entry.
If you plan to come by car, make sure to include your license plate number so Cisco can register you for access to their parking garage. You’ll also receive an email with a personal wireless account for the event.
https://forms.gle/TBDzqxk4h6JHTJyw6
Designing for the Real World: The Next Phase of GenAI
As Generative AI moves from experiments to real-world use, teams are starting to face deeper challenges — from securing data in retrieval pipelines to figuring out when a GenAI app is actually ready for production.
This event brings together two talks focused on practical strategies for designing GenAI systems that are secure, well-evaluated, and ready for the complexity of real use cases.
We’ll cover:
• Why authorization is critical in Retrieval-Augmented Generation (RAG) pipelines
• Techniques for pre-filtering and post-filtering in vector databases
• A practical demo using Pinecone, LangChain, OpenAI, and SpiceDB
• What makes a GenAI app “ready” for production
• How evaluation frameworks can improve quality and reduce time-to-market
• Lessons from real-world teams building and deploying GenAI systems
Speaker 1:
Sohan Maheshwar (Lead Developer Advocate, AuthZed)
Sohan is a seasoned DevRel expert with experience at Amazon, Fermyon, and Gupshup. Based in the Netherlands, he now leads developer advocacy at AuthZed, focusing on infrastructure and security in modern software systems.
Talk: Securing RAG Pipelines with Fine Grained Authorization
As GenAI moves into enterprise environments, securing sensitive data in RAG pipelines becomes critical — especially in regulated industries.
In this talk, Sohan will explain why fine-grained authorization is essential for protecting data, and how modern permission systems make this possible. He’ll walk through techniques like prefiltering and post-filtering in vector databases, with a live demo using Pinecone, LangChain, OpenAI, and SpiceDB.
Speaker 2:
Carlo Provinciali (Technical Product Manager, Booking.com)
Carlo is a former data scientist now working as a Technical Product Manager at Booking.com, where he helps teams design and ship GenAI applications that work at scale.
Talk: Evals is All You Need? Mastering GenAI Application Quality
Building with LLMs is unpredictable — outputs vary, behavior shifts, and it's hard to know when a GenAI app is truly ready for production.
In this talk, Carlo will explore how evaluation frameworks can bring structure and confidence to the GenAI development process. He’ll show how the right eval strategy helps teams consistently measure quality, make better decisions, and reduce time-to-market.
Agenda:
18:00 — Arrival, food & drinks
18:30 — Talk #1 | Sohan Maheshwar
19:15 — Short break
19:30 — Talk #2 | Carlo Provinciali
20:15 — Open conversation, networking & more drinks
21:00 — Wrapping up
What to bring:
Just curiosity and questions. If you're working on GenAI infrastructure, evaluation tooling, or secure platform design, we’d love to hear your thoughts.
Who is this for:
Platform engineers, AI infra builders, technical product managers, and engineering leaders navigating the shift toward building reliable GenAI systems in production.
Where to find us:
Cisco HQ, Haarlerbergweg 13-19, 1101 CG Amsterdam-Zuidoost.
