WeAreDevelopers Logo WeAreDevelopers
Sign in
For You LIVE
World Congress 2026 Europe • Jul 10, 2026 • Session details

Localized Open Models in Production: What Builders Need to Know

Ankit Patel , Pierre-Louis Cedoz , Jamie Madden , Stephen Batifol

A less capable model with a highly engineered workflow consistently outperforms a massive frontier model. Learn how to quantize weights and deploy enterprise-ready, localized AI on standard hardware.

Pause
Mute Enter Fullscreen
#1 about 2 min

Defining local open AI models for production environments

Owning and executing model weights directly differentiates local deployment from standard cloud provider APIs.

#2 about 2 min

Leveraging quantization techniques to scale local hardware execution

Compressing models with advanced quantization formats like FP4 enables massive parameter sets on consumer GPUs.

#3 about 4 min

Implementing enterprise governance and logging for multi-agent workflows

Capturing comprehensive system logs and human-in-the-loop approvals satisfies strict compliance and workflow auditing requirements.

#4 about 2 min

Defining execution environments for browser and computer use agents

Providing secure desktop or web sandboxes mitigates risk when computer-use agents interact with unstructured environments.

#5 about 5 min

Evaluating subjective generative media with artificial and human judges

Combining vision-language model scoring with manual oversight bridges the difficulty of evaluating nuanced aesthetic outputs.

#6 about 4 min

Optimizing execution costs and inference speeds with localized endpoints

Post-training smaller deterministic models for localized execution prevents escalating API costs and excessive inference latency.

#7 about 5 min

Automating workloads using scheduled agents and continuous model fine-tuning

Deploying long-running background agents to retrieve data and refine models eliminates tedious infrastructure management tasks.

#8 about 3 min

Combating workflow drift and integrating underlying infrastructure updates

Consistently auditing agent outputs and applying underlying driver enhancements guarantees long-term reliability and compounded performance gains.

#9 about 5 min

Prototyping multi-step workflows quickly within accessible model playgrounds

Verifying foundational model capabilities in web-based playgrounds prevents wasted engineering effort on overly complex workflow harnesses.

#10 about 3 min

Simulating physics and complex realities using emergent world models

Deploying architectures that intrinsically understand physical logic drives advanced video simulation and autonomous robotic control.

Matching moments

2:17 min

Transitioning generative AI from experimentation to production

Adolf Hohl Adolf Hohl · World Congress 2024

1:32 min

Constructing scalable AI factory stack blueprint architectures

Anshul Jindal Anshul Jindal +1 · World Congress 2026 Europe

2:02 min

Shifting focus from isolated models to enterprise AI systems

Mohak Chadha Mohak Chadha · World Congress 2026 Europe

1:09 min

Deploying algorithms and AI models to edge production

Irina Terekhova Irina Terekhova · World Congress 2026 Europe

1:48 min

Identifying barriers to enterprise generative AI production environments

Stephan Gillich Stephan Gillich · World Congress 2024

4:44 min

Introduction to prototyping and building practical AI applications

Krzysztof Cieślak Krzysztof Cieślak · World Congress 2024

More from World Congress 2026 Europe

Building the Nervous System of AI - Michael Kagan (NVIDIA) thumbnail
29:17
3 months ago

Building the Nervous System of AI - Michael Kagan (NVIDIA)

Michael Kagan, Sead Ahmetović

Five Things in Tech that Matter Now - WeAreDevelopers World Congress 2026 Closing Keynote thumbnail
19:41
3 months ago

Five Things in Tech that Matter Now - WeAreDevelopers World Congress 2026 Closing Keynote

Christian Heilmann

The Sound of Your Secrets: Teaching Your Model to Spy, So You Can Learn to Defend thumbnail
32:06
3 months ago

The Sound of Your Secrets: Teaching Your Model to Spy, So You Can Learn to Defend

David vonThenen

Super scaling for the Super Bowl: How to survive 30 million users hitting your backend in 30 minutes thumbnail
35:13
3 months ago

Super scaling for the Super Bowl: How to survive 30 million users hitting your backend in 30 minutes

Irina Branovic

Running Secure Life Science Research at Scale using Hybrid GPU HPC and Kubernetes 🧬 thumbnail
24:23
3 months ago

Running Secure Life Science Research at Scale using Hybrid GPU HPC and Kubernetes 🧬

Jeremy Murray

Angstfreude - AI and Corporate Culture - The Thrill and the Threat thumbnail
30:26
3 months ago

Angstfreude - AI and Corporate Culture - The Thrill and the Threat

Axel Ebert, Karin Krobath

Related videos

AI That Fits Your Business, Not the Other Way Around thumbnail
25:56
World Congress 2026 Europe • 3 months ago

AI That Fits Your Business, Not the Other Way Around

Anshul Jindal, Cansu Kavili Örnek

RTX AI PC: Developing local and edge AI applications thumbnail
26:06
World Congress 2026 Europe • 3 months ago

RTX AI PC: Developing local and edge AI applications

Joerg Krall

Running AI at Scale: The Secret Ingredients thumbnail
30:04
World Congress 2026 Europe • 3 months ago

Running AI at Scale: The Secret Ingredients

Boris Hecker, Max Tschochohei, Peter Kürpick, Tomislav Tipurić

The Last Mile of AI: From Prototype to Production thumbnail
28:23
World Congress 2026 Europe • 3 months ago

The Last Mile of AI: From Prototype to Production

Clemens Wasner, Holger Hammel, Martin Remmelgas, Max Jacobson, Marius Cosareanu

Open Source AI, To Foundation Models and Beyond thumbnail
31:05
World Congress 2025 • about 1 year ago

Open Source AI, To Foundation Models and Beyond

Andreas Blattmann, Ankit Patel, Lucie-Aimée Kaffee, Matt White, Philipp Schmid

Your Next AI Needs 10,000 GPUs. Now What? thumbnail
29:52
World Congress 2025 • about 1 year ago

Your Next AI Needs 10,000 GPUs. Now What?

Anshul Jindal, Martin Piercy

Related articles

From learning to earning

Jobs that call for the skills explored in this talk.