Your Data. Your Models. Your Cloud: Bringing AI Inference In-House
How GoodData and Superlinked are approaching private, sovereign AI with smaller models and infrastructure you control.
Your customers want AI features. Your security team wants to know where the data goes. And your engineering team needs a way to deliver reliable AI without depending entirely on external model providers.
So what does it actually take to bring inference in-house?
Join GoodData and Superlinked for a practical look at building and deploying AI within your own infrastructure.
Zdeněk Musil (CTO, AI at GoodData) will share GoodData's journey towards private inference, including the security requirements driving the shift, the challenges of evaluating smaller models, and the lessons learned along the way.
We'll explore how enterprises can combine specialised small models with larger LLMs, using confidence-based routing to balance accuracy, performance and cost while keeping sensitive workloads within their security boundaries.
Then Daniel Svonava (CEO of Superlinked) will take us beneath the hood with a 15-minute technical demo of self-hosted inference. Expect architecture diagrams, code and a walkthrough of how multiple specialised models can run together in your own cloud using Superlinked's Inference Engine (SIE).
Demetrios Brinkmann will host the discussion, challenging the speakers on the practical realities of deploying and operating AI in enterprise environments.
What you'll learn
Data security and sovereignty: How to run inference in your own cloud or private infrastructure, with control over where sensitive data is processed.
Choosing the right models: How smaller, task-specific models can handle classification, guardrails and decision-making, and where larger LLMs are still needed.
Making private AI reliable: How to evaluate model accuracy, handle failures and maintain quality as models and workloads evolve.
Building the infrastructure: A technical walkthrough of the architecture, deployment and model-serving stack behind in-house inference.
Controlling costs and dependencies: How to reduce reliance on external AI APIs and make inference costs more predictable.
If you're a CTO, AI infrastructure lead or platform engineer navigating enterprise security requirements, data sovereignty or the transition to self-hosted AI, this session will show you what bringing inference in-house actually involves.
Real engineering challenges. Real architecture. AI infrastructure under your control.