Marco González
Red Hat, Sr. Software Engineer
Tokyo, Japan
Actions
Principal Engineer at Red Hat with 13+ years of global experience in 3G, 4G, and 5G network integration. US patent creator and Telco Automation expert. Active CNCF Program Committee member and speaker at KubeCon, Kubedays, and CNCF events. Contributor to Open5GCore, Kepler, and kube-compare. Currently focused on MCP ecosystem architecture and production standards for AI tool interoperability on Kubernetes.
Badges
Area of Expertise
Topics
Exploring Kepler's Next Chapter: Achieving Cloud-Native Sustainability with MCP Integration
The Kepler open-source project is vital for monitoring energy consumption in Kubernetes. However, its raw and complex data often exists in a vacuum that Enterprises take time to interpret and use. This presentation introduces the next evolution: integrating Kepler with the Model Context Protocol (MCP). This transforms Kepler into an interactive, queryable data source, allowing developers to contextualize energy metrics in real time. I'll explore how this new approach, empowers developers to not only observe power consumption but to compare it against local regulatory frameworks, such us MIC in Japan, and international standards. The session culminates in a live demo where we will use MCP to query a high-CPU workload, instantly validating its energy profile against predefined compliance targets. This powerful combination dramatically accelerates the time-to-market for verified green and compliant cloud-native applications.
Millisecond AI: LLM Inference at the 5G Edge
With 5G’s rollout, delivering AI services in milliseconds has never been more critical.
This session shows how to deploy large language models (LLMs) on edge infrastructure leveraging Open5GS for a virtualized 5G core and Ollama for local inference to slash latency and cloud expenses.
We’ll walk through a production-grade architecture and demo a simulated 5G device calling an edge-hosted AI endpoint. You’ll learn how to optimize workload placement, enable CPU-only inference, and balance reliability with resource constraints when integrating AI into telecom networks.
By the end, you’ll have a practical blueprint for bringing real-time intelligence to users and unlocking new edge-driven innovation in the AI & Data Innovations track.
Launching SynapseAI: A Lean, Trustworthy Framework for Autonomous AI Agents at Scale
As autonomous AI agents gain traction across industries, frameworks like Auto-GPT and CrewAI have showcased their potential but also exposed critical limitations: uncontrolled token usage, narrow domain adaptability, and lack of enterprise-grade safety controls.
In this session, we will introduce SynapseAI, a new open-source framework designed to build cost-efficient, scalable, and safe AI agents that go beyond experimental use. With a modular multi-agent architecture, SynapseAI enables intelligent delegation through specialized roles (Planner, Executor, Verifier, Auditor) and integrates dynamic model selection and token budgeting to reduce cost and carbon footprint.
Its core innovation lies in the Trust Layer, a human-aligned safety system that enables approval gates, transparent logging, and rollback-safe actions. Designed for real-world use across industries, SynapseAI includes plug-and-play integrations with enterprise APIs and domain-adaptive plugins for DevOps, finance, logistics, and healthcare.
This session will walk attendees through the motivation, design, and use cases of SynapseAI, including a live DevOps demo and practical steps to get started. Whether you're building internal automation or scaling customer-facing agents, SynapseAI offers a reliable foundation.
From Misconfigured to Production-Ready: Open Source RAN DU Validation for AI-Native Telco
Anyone deploying AI-native RAN workloads knows the first challenge is simply getting the cluster right. As the industry moves toward autonomous networks, what comes next at the edge?
Rigid, manually tuned RAN DU configurations were never built to co-host AI inference. When a single edge node must handle real-time L1/L2 RAN processing alongside reasoning models for fault isolation and network optimization, the scheduling policy and validation pipeline both need to change.
This talk guides Telco attendees through building an AI-native RAN DU stack using OKD: CPU pinning, NUMA alignment, PREEMPT_RT kernel tuning, and GPU accelerator integration for co-located Edge AI inference. We show how unified CPU+GPU memory eliminates the bottleneck separating RAN processing from AI workloads.
The session closes with a live demo on a real lab cluster: we introduce a topology manager policy misconfig and a CPU isolation parameter error, two of the most common silent failures in RAN DU deployments, then run kube-compare to surface both with precise remediation guidance. The audience watches the cluster go from failing validation to a clean pass in real time, over a live remote connection.
With Great Networks Come Great Carbon Savings: OpenSource + AI for RAN Optimization
Telecom networks, especially with 5G expansion, are major energy consumers. What if we could cut energy use and carbon emissions without compromising performance? In this session, we’ll show how Kubernetes and OpenSource tools can transform Radio Access Networks (RAN) into energy-efficient systems.
We’ll walk through deploying AI/ML pipelines to predict network traffic and optimize energy in real-time. Using Kubernetes for scalability, Prometheus and Grafana for monitoring, and open-source tools for automation, we’ll demonstrate how these technologies come together to tackle energy challenges.
The session ends with a live demo, starting with a RAN simulator, integrating AI workloads, and visualizing energy-saving results. You’ll see the power of OpenSource in action and leave with practical tips, reusable code, and a clear vision for making telecom networks more sustainable.
Next-Level 5G Core Deployments: Harnessing eBPF's Capabilities for Performance and Security
This session will focus on the power of applying eBPF to 5G Core deployments over Kubernetes.
We aim to examine and discuss how eBPF can improve network performance and security and contribute to the evolution of existing 5G technologies.
Our presentation will start with an overview of eBPF – a revolutionary technology that allows for dynamic tracing of the kernel, enhancing network routing, load balancing, and security. We will discuss challenges for current 5G deployment strategies and how eBPF can provide practical solutions, emphasizing its potential benefits and relevance in the 5G context. Then we will transition into a detailed exploration of deploying a CNI with eBPF in a 5GCore solution over Kubernetes.
The last part of the session will conclude with a real-world demonstration of deploying LTERAN and establishing end-to-end communication, showcasing the practical implications and benefits of using eBPF in 5G deployment strategies and some takeaways.
Mission Possible: Forging a New Paradigm in 5G Deployment through CDK and AI Issue Navigation
This talk will outline the deployment and fine-tuning of the 5G core. Our approach utilizes customized CDK constructs and LLMs to establish a fully operational Edge 5G core network.
In the first part of the session, we will showcase a new way of combining LLM with our existing 5G core network, which will be deployed using CDK construct. We will then use data from logs to continuously feed and update the training pipeline. This setup enables open-source LLMs to quickly adapt and provide on-the-spot, practical insights.
The last part of the session will conclude with a real-world demonstration where the LLM successfully collects logs in real-time including the 5G codebase, to serve as a training dataset to diagnose and rectify any challenging error in this Edge 5G network, showcasing its capability to analyze and suggest corrective actions effectively. This approach not only enhances operational efficiency but also optimizes response strategies in 5G telecommunications infrastructure.
Mission Possible: Forging a New Paradigm in 5G Deployment through CDK and AI Issue Navigation
This talk will outline the deployment and fine-tuning of the 5G core. Our approach utilizes customized CDK constructs and LLMs to establish a fully operational Edge 5G core network.
In the first part of the session, we will showcase a new way of combining LLM with our existing 5G core network, which will be deployed using CDK construct. We will then use data from logs to continuously feed and update the training pipeline. This setup enables open-source LLMs to quickly adapt and provide on-the-spot, practical insights.
The last part of the session will conclude with a real-world demonstration where the LLM successfully collects logs in real-time including the 5G codebase, to serve as a training dataset to diagnose and rectify any challenging error in this Edge 5G network, showcasing its capability to analyze and suggest corrective actions effectively. This approach not only enhances operational efficiency but also optimizes response strategies in 5G telecommunications infrastructure.
A Paradigm Shift in 5G Deployment: The Promise of eBPF for Open-source Core Performance and Security
This session will focus on the power of applying eBPF to 5G Core deployments over Kubernetes.
We aim to prove and discuss how eBPF can improve network performance and security and contribute to the evolution of existing 5G technologies.
Our presentation will start with an overview of eBPF – a revolutionary technology that allows for dynamic tracing of the kernel, enhancing network routing, load balancing, and security. We will discuss challenges for current 5G deployment strategies and how eBPF can provide practical solutions, emphasizing its potential benefits and relevance in the 5G context.
Then we will transition into a detailed exploration of deploying a CNI with eBPF in a 5GCore solution over Kubernetes.
The last part of the session will conclude with a real-world demonstration of deploying LTERAN and establishing end-to-end communication, showcasing the practical implications and benefits of using eBPF in 5G deployment strategies and some takeaways.
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