Articles from Source: Red-Hat-Developer-Blog

Red Hat Developer Hub: Preventing compliance violations with AI coding agents

2026-09-15 13:17
πŸš€ AI coding agents are transforming service development by generating entire services and scaffolding applications quickly. However, they may not align with your organization’s specific compliance rules or existing services. Understanding these gaps is crucial to ensure compliance and efficiency. Explore how Red Hat Developer Hub addresses these challenges! #AICoding #Compliance #RedHat #SoftwareDevelopment #TechTrends
Source: Red Hat Developer Blog
Evan Shortiss, Ben Wilcock

Understanding W8A8 INT8 LLM quantization: Accuracy and performance results

2026-09-14 13:01
πŸš€ Exciting progress in model quantization! The article discusses how the Llama 3.1 8B Instruct model was compressed from 14.9 GB to 8.0 GB using 8-bit integer (INT8) W8A8 quantization techniques, including SmoothQuant and GPTQ. This approach maintains accuracy while enhancing performance. #AI #Quantization #MachineLearning #DataScience #ModelOptimization
Source: Red Hat Developer Blog
Sana Fayyaz

Python 3.14 free-threaded build is now available in RHEL

2026-09-14 13:01
πŸš€ Great news for developers! Python 3.14 is now available in a free-threaded build for Red Hat Enterprise Linux 9.8 and 10.2. This new version allows for full parallel CPU execution, enhancing performance. You can find both the standard and free-threaded variants in the Red Hat CodeReady Linux Builder repositories. This marks the first official support for the free-threaded build. #Python #RedHat #Development #RHEL #TechNews
Source: Red Hat Developer Blog
LumΓ­r Balhar

Unlock a LUKS root over SSH on Fedora and Red Hat Enterprise Linux

2026-09-11 13:01
πŸ”’ Is your encrypted server stuck in initramfs? This article explains how to unlock a LUKS root over SSH on Fedora and RHEL. It details three commands and a kernel argument that enable NetworkManager and sshd in the initramfs environment. This allows remote unlocking of your encrypted root, ensuring smoother server management. For more insights, check out the full post! #Linux #Fedora #RHEL #ServerManagement #Encryption
Source: Red Hat Developer Blog
Josephine Pfeiffer

Bringing custom knowledge to agents with AutoRAG

2026-09-11 13:01
Unlock the potential of large language models (LLMs) with AutoRAG! These models rely on training data to understand context, including specific corporate jargon. Without the right data, they may misinterpret information, leading to inaccuracies. AutoRAG aims to enhance LLMs by providing tailored knowledge, improving their response quality. #AI #AutoRAG #LanguageModels #TechInnovation #DataScience πŸ€–πŸ“Š
Source: Red Hat Developer Blog
Diego Alvarez Ponce, Diego Garcia Perez

Red Hat edge platforms: Choosing the right one for your use case

2026-09-11 13:01
Choosing the right edge platform is crucial for organizations. Options vary widely, from basic single-board computers to complex Kubernetes clusters with advanced features. Selecting an inadequate platform can lead to challenges as needs evolve. Understanding your requirements is key to making an informed decision. πŸ”§πŸ’»πŸŒ #EdgeComputing #RedHat #Kubernetes #TechDecisions #CloudSolutions
Source: Red Hat Developer Blog
Daniel Froehlich

From token consumer to token provider: Building your org's AI API

2026-09-10 13:01
AI implementation is proving costly for organizations, as many have opted to use external AI providers for their workflows. Initially, this approach seemed effective due to accessible APIs and reliable models. However, unexpected expenses have raised concerns among businesses. Exploring the shift from being token consumers to becoming token providers could help organizations manage costs better. πŸŒπŸ’‘ #AI #BusinessStrategy #TechTrends #Innovation
Source: Red Hat Developer Blog
Markell Rawls

Deploy NeMo Guardrails on Red Hat OpenShift AI

2026-09-09 15:05
In part 3 of the series on local guardrail development, the focus is on deploying NeMo Guardrails on Red Hat OpenShift AI. The previous articles covered designing guardrail configurations and conducting manual testing. The second part emphasized rigorous testing against large-scale risk datasets. This series offers valuable insights for developers. πŸ“ŠπŸ”§ #NeMoGuardrails #RedHat #OpenShiftAI #TechDevelopment #DataSafety
Source: Red Hat Developer Blog
Rob Geada

Use a local and open source code assistant

2026-09-09 14:01
Explore the benefits of using local and open source coding assistants! Many popular AI coding tools lack transparency or compromise your data privacy by relying on cloud processing. For those seeking alternatives, consider integrating an open coding assistant with an open source IDE. Stay in control of your projects while enhancing your coding experience! πŸ’»πŸ” #OpenSource #Coding #DataPrivacy #AI #DeveloperTools
Source: Red Hat Developer Blog
Seth Kenlon

Personalize your product's text-to-speech voice for any language: Fine-tuning with Kubeflow Trainer on Red Hat OpenShift AI

2026-09-09 03:32
πŸ“Š A recent CSA Research survey highlights that 76% of consumers prefer to buy in their native language, and 40% won’t purchase in another. This preference extends to how products communicate verbally. Explore how to personalize text-to-speech voices using Kubeflow Trainer on Red Hat OpenShift AI. #TextToSpeech #LanguagePreference #AI #RedHat #CustomerExperience
Source: Red Hat Developer Blog
Dmytro Hryshchenko, Abhijeet Dhumal

Optimize vLLM speculative decoding with FastMTP heads

2026-09-08 14:20
Large language models (LLMs) face challenges with autoregressive decoding, as they rely heavily on memory bandwidth. Each token requires a full forward pass over billions of parameters, leading to inefficiencies. The article discusses the MTP training objective, which helps models like DeepSeek and Qwen predict multiple future tokens, enhancing data efficiency and quality. For more insights, check out the full article! πŸ“ˆπŸ’‘ #LLM #DataEfficiency #MachineLearning #FastMTP #RedHatDeveloper
Source: Red Hat Developer Blog
Rahul Tuli

Understanding W8A8 INT8 LLM quantization: Half the size, better performance, same accuracy

2026-09-07 07:01
Large language models like Llama 3.1 8B require significant GPU memory, around 15 GB when using Bfloat16 (BF16) precision. This includes storage for 8 billion parameters and additional memory for key-value caches and activations during inference. W8A8 INT8 quantization offers a solution by reducing model size while maintaining performance and accuracy. πŸ”πŸ’‘πŸ“‰ #MachineLearning #AI #ModelOptimization #Quantization
Source: Red Hat Developer Blog
Sana Fayyaz

Red Hat Developer Hub software template authoring with rhdh-templates

2026-09-07 07:01
πŸš€ Red Hat Developer Hub offers a new approach to software template authoring with rhdh-templates. Platform engineers and developers often face challenges with Nunjucks syntax and locating YAML files. The process can lead to errors that are only identified after rendering in the developer portal. AI tools may assist with generic YAML, but they do not support the specific conventions of Red Hat Developer Hub and Backstage. #RedHat #DeveloperHub #YAML #SoftwareTemplates #TechNews
Source: Red Hat Developer Blog
Kashish Mittal

Reference architecture for HA scanning with Red Hat Advanced Cluster Security for Kubernetes

2026-09-07 03:00
πŸ” Red Hat Advanced Cluster Security for Kubernetes offers an image scanning API crucial for CI/CD pipelines assessing vulnerabilities. 🚫 However, upgrades and restarts can cause downtime, disrupting essential build processes. This article outlines a reference architecture that resolves this issue by utilizing two Central service instances along with a client-side failover mechanism. #Kubernetes #RedHat #DevOps #Cybersecurity #HAScanning
Source: Red Hat Developer Blog
Johannes Malsam

From incident to remediation: Building an AI-driven AIOps workflow with Red Hat Ansible Automation Platform

2026-09-05 03:01
Unlocking the potential of AI in operations is crucial for effective incident management. In a recent article, a basic AIOps workflow was developed using AWS EC2, Red Hat Enterprise Linux, Red Hat Ansible Automation Platform, and Claude Code. The focus is on transforming insights from incidents into safe, repeatable actions. Explore how technology can streamline operations! πŸŒπŸ’» #AIOps #RedHat #Automation #TechTrends #IncidentManagement
Source: Red Hat Developer Blog
Neha Chugh

Build AI-assisted firewall workflows with Ansible Automation Platform

2026-09-04 07:01
Unlock the power of automation with Ansible! πŸ”’βœ¨ The article discusses how to integrate AI-assisted workflows into firewall management. It highlights the ability to manage access requests, such as allowing app-client-01 to connect to the database on port 5432. Explore how Ansible can streamline these processes effectively. #Ansible #Automation #Cybersecurity #AI #FirewallManagement
Source: Red Hat Developer Blog
Michal Zdyb

Configure admission fair sharing in Red Hat build of Kueue 1.4

2026-09-04 07:01
In shared AI and high-performance computing clusters, managing limited hardware resources like GPUs and CPUs is crucial for OpenShift platform engineers. The FIFO admission rule can lead to long wait times for later workloads, impacting service predictability. Red Hat build of Kueue 1.4 introduces two methods for ensuring fairness: fair sharing-based preemption and admission fair sharing. Learn more about optimizing resource management! πŸš€πŸ’» #OpenShift #Kueue #ResourceManagement #AI...
Source: Red Hat Developer Blog
Maysa De Macedo Souza

Speeding up LLM inference with P-EAGLE in vLLM Speculators

2026-09-03 17:56
πŸš€ Amazon has introduced P-EAGLE, a new speculative decoding algorithm designed to enhance LLM inference. This innovation builds on EAGLE-3 by incorporating parallel drafting, which aims to improve efficiency in processing. Discover how P-EAGLE is set to advance speculative decoding technology in the latest update. #TechInnovation #MachineLearning #Amazon #SpeculativeDecoding #AI
Source: Red Hat Developer Blog
Helen Zhao, Megan Flynn, Dipika Sikka

Evaluating LLM guardrail configs locally with EvalHub

2026-09-03 16:58
In part 2 of the series on local guardrail development, the focus is on evaluating LLM guardrail configurations using EvalHub. The article expands on the design and development process discussed in the first part, emphasizing the importance of manual testing to ensure effectiveness. Stay tuned for further insights on enhancing local configurations! πŸ”πŸ’» #EvalHub #LLM #GuardrailDevelopment #RedHat #TechInsights
Source: Red Hat Developer Blog
Rob Geada

Evaluate AI agents with IBM CLEAR & EvalHub on OpenShift AI

2026-09-03 07:01
πŸš€ Adopting eval-driven development is crucial for reliable AI agents. However, implementation can be challenging. Even agents that perform well in basic scenarios may face subtle issues in production. These include misusing tools, straying off-topic, or providing plausible but incorrect responses. Unit tests often miss these failures, highlighting the need for better evaluation methods. #AI #EvalDrivenDevelopment #IBM #OpenShift #TechInsights
Source: Red Hat Developer Blog
Hema Veeradhi, Surya Pathak

Extend Layer 2 networks into Red Hat OpenShift Virtualization with BGP and EVPN

2026-09-03 03:01
πŸš€ Migrating to Red Hat OpenShift Virtualization can modernize your infrastructure without disrupting existing network settings. However, traditional systems often face challenges in maintaining connectivity for imported VMs and ensuring multi-tenant isolation. The article highlights solutions using BGP and EVPN to streamline these processes. #RedHat #OpenShift #Virtualization #Networking #BGP #EVPN
Source: Red Hat Developer Blog
Miguel Duarte de Mora Barroso, Valentino Uberti

LoRA backdoor threat: How OpenShift AI mitigates the risk

2026-09-02 07:01
A recent article discusses the risks associated with deploying productivity assistants in pharmaceutical research. A researcher utilized a trusted open-weight model, enhanced with a low-rank adaptation (LoRA) adapter for specific lab needs. While the main model is reliable, the adapter is often overlooked, raising concerns about potential vulnerabilities. OpenShift AI offers strategies to mitigate these risks and ensure safer implementation. πŸ”πŸ’‘ #AI #Pharmaceuticals #Cybersecurity #OpenSource...
Source: Red Hat Developer Blog
Mike Hepburn

LLM quantization guide: How to do it, and how it helps

2026-09-02 03:16
πŸš€ Model sizes are rapidly increasing, doubling each year, while GPU memory struggles to keep pace. πŸ“ˆ The article discusses LLM quantization, a technique that enables efficient deployment of large models on accessible hardware. This is essential for serving multiple users effectively. πŸ’‘ Learn how quantization can help in managing resource limitations while maximizing performance. #LLM #Quantization #MachineLearning #AI #TechTrends
Source: Red Hat Developer Blog
Cedric Clyburn

Orchestrate production RAG with OpenShift AI

2026-09-01 07:01
Discover how to enhance your streaming retrieval-augmented generation (RAG) pipeline with OpenShift AI. The previous article outlined a single Ray Data script that handles parsing, chunking, embedding, and writing to Milvus. However, this monolithic approach has a drawback: if a parsing error occurs, you must restart the entire process. Learn more about optimizing your RAG workflow! πŸš€πŸ’» #OpenShiftAI #DataPipeline #RAG #TechInnovation #RedHatDeveloper
Source: Red Hat Developer Blog
Ana Biazetti, Saad Zaher

Developing LLM guardrail configs locally with NeMo Guardrails

2026-09-01 03:01
πŸš€ Large language models (LLMs) offer great potential, but deploying them comes with risks like prompt injection and data leakage. 🀝 Red Hat has teamed up with NVIDIA to create NeMo Guardrails, an open source framework designed to add safety measures to LLM applications. πŸ”§ This solution enables developers to implement programmable guardrails, enhancing the security of their AI systems. #AI #MachineLearning #DataSecurity #NeMoGuardrails #OpenSource
Source: Red Hat Developer Blog
Rob Geada

Red Hat OpenShift autoscaling using MachineSet autoscaler with KEDA

2026-09-01 03:01
Discover the latest in Red Hat OpenShift scaling solutions! This article explores the MachineSet Autoscaler with KEDA, a metrics-driven method for scaling individual MachineSets based on external or custom signals. Previously, we discussed the Cluster Autoscaler, a built-in, Kubernetes-native approach. Learn how these options can enhance your compute infrastructure! πŸ’»πŸ“ˆ #RedHat #OpenShift #KEDA #Autoscaling #CloudComputing
Source: Red Hat Developer Blog
Ramon Gordillo Gutierrez, Jose Ortiz Padilla

AI-powered multicluster management: Querying fleet health with OpenShift Lightspeed and Red Hat Advanced Cluster Management

2026-08-31 07:16
Managing large-scale Kubernetes environments presents significant challenges, especially in maintaining visibility and troubleshooting across multiple clusters. As the number of managed clusters increases, platform engineers face a higher cognitive load, which can slow response times during critical incidents. OpenShift Lightspeed and Red Hat Advanced Cluster Management offer solutions to help manage fleet health effectively. πŸ”§πŸ’»πŸš€ #Kubernetes #OpenShift #RedHat #CloudManagement #TechSolutions
Source: Red Hat Developer Blog
Diego Alvarez Ponce, Luiz Bernardo Levenhagen

Stop wasting GPU allocation in Kubernetes with GPU-pruner

2026-08-31 03:01
Are idle GPUs costing you? πŸ–₯οΈπŸ’Έ In Kubernetes, wasted GPU resources can lead to significant expenses. Users often find GPUs allocated for days with minimal activity, resulting in unnecessary charges. The article introduces GPU-pruner, a tool designed to address this issue by optimizing GPU allocation. Learn more about how to manage your resources effectively! #Kubernetes #GPUManagement #CloudComputing #TechSolutions #ResourceOptimization
Source: Red Hat Developer Blog
Fahim Uddin

Self-service backup for VMs and containers on OpenShift: No cluster-admin required

2026-08-28 08:16
Running a multi-tenant Red Hat OpenShift cluster can be challenging. Currently, every backup and restore with Velero needs cluster-admin privileges. This limits application teams from managing their own workloads effectively. They often rely on the platform team, creating delays and bottlenecks in the process. A new approach to self-service backup may help streamline these operations. πŸ”„πŸ’»πŸ’‘ #OpenShift #Backup #CloudComputing #DevOps #RedHat
Source: Red Hat Developer Blog
Michal Pryc, Shubham Dilip Pampattiwar

Building hermetic notebook images for Open Data Hub and Red Hat OpenShift AI

2026-08-28 03:01
πŸš€ Building reliable notebook images for Open Data Hub and Red Hat OpenShift AI has been streamlined. Key to this process is the concept of hermetic builds, where nothing is downloaded during the build. Dependencies are pre-fetched and installed from a local cache, ensuring consistency and eliminating network issues. This approach works seamlessly across different environments, including laptops and GitHub Actions. πŸ”§πŸ’» #OpenDataHub #RedHat #DevOps #Containerization #AI
Source: Red Hat Developer Blog
Vath Sok

llm-d flow control: Priority queuing for shared GPU inference

2026-08-27 03:01
πŸš€ llm-d flow control introduces priority-aware admission and tenant fairness scheduling for GPUs. This enhancement allows platform teams to effectively manage mixed workloads on shared model pools. Now available in Red Hat AI Inference 3.5, this feature aims to optimize GPU resource utilization. #RedHat #AIInference #GPU #TechUpdates #MachineLearning
Source: Red Hat Developer Blog
Alexa Griffith, Rishabh Saini

How AI observability works with MLflow

2026-08-26 09:16
Understanding AI observability is crucial, especially since AI responses can be inaccurate. The article explores how MLflow is utilized to enhance AI observability, ensuring better monitoring and evaluation of AI models. For those interested in improving AI accuracy, this piece provides valuable insights. πŸ€–πŸ“Š #AI #MLflow #DataScience #TechInsights
Source: Red Hat Developer Blog
Cedric Clyburn

Try the Ansible playbook generation lab with Gemini and OpenAI

2026-08-26 07:01
πŸš€ Red Hat's Ansible Automation Platform has enhanced its VS Code extension since 2021. Key features include automatic linting and simplified documentation access. A standout capability is the generation of Ansible playbooks and roles using AI language models. Explore the Ansible playbook generation lab with Gemini and OpenAI! #RedHat #Ansible #Automation #AI #VSCode
Source: Red Hat Developer Blog
clobner

Automating Red Hat OpenShift AI installations with Helm and GitOps

2026-08-26 03:01
Discover a new method for automating Red Hat OpenShift AI installations using Helm and GitOps. This article builds on previous insights about managing OpenShift AI dependencies and presents a more streamlined approach for deployment. Learn how this can enhance your workflow! πŸš€πŸ”§ #RedHat #OpenShiftAI #GitOps #Helm #Automation
Source: Red Hat Developer Blog
Davide Bianchi, Andrea Tarocchi

GRPO fine-tuning on Red Hat OpenShift AI: Reinforcement learning from verifiable rewards with Training Hub

2026-08-26 03:00
Large language models are great for conversation, but enterprise needs often require structured reasoning, like generating SQL or JSON. Supervised fine-tuning improves these skills but can be costly and time-consuming to curate data. Explore how GRPO fine-tuning on Red Hat OpenShift AI addresses these challenges with reinforcement learning. #AI #RedHat #MachineLearning #DataScience #EnterpriseSolutions πŸ€–πŸ“Š
Source: Red Hat Developer Blog
Fiona Waters

GRPO fine-tuning on Red Hat OpenShift AI: Reinforcement learning from verifiable rewards with Training Hub

2026-08-25 03:16
Explore the latest advancements in GRPO fine-tuning on Red Hat OpenShift AI. This article discusses how reinforcement learning can utilize verifiable rewards to enhance training efficiency. Key insights include the role of the Training Hub in facilitating this process and its implications for developers in AI. Stay informed about the future of AI development! πŸ€–πŸ“Š #RedHat #OpenShiftAI #ReinforcementLearning #AI #TrainingHub
Source: Red Hat Developer Blog
Fiona Waters

Red Hat OpenShift autoscaling with Cluster Autoscaler

2026-08-25 03:16
πŸš€ Explore the capabilities of the Red Hat OpenShift Cluster Autoscaler! This built-in tool automatically adjusts your cluster size based on workload demands. It integrates seamlessly with the Red Hat OpenShift machine API, using two key resources: ClusterAutoscalerAPI and MachineAutoscalerAPI. Learn how it handles scale-up and scale-down operations, ensuring efficient resource management. Check out the article for practical examples on Microsoft Azure! #RedHat #OpenShift #Autoscaler...
Source: Red Hat Developer Blog
Ramon Gordillo Gutierrez, Jose Ortiz Padilla

Run LoRA fine-tuning on Red Hat OpenShift AI with Ray

2026-08-24 07:01
πŸš€ Ray is a powerful open-source framework for scaling AI workloads, now integrated into Red Hat OpenShift AI. With the inclusion of the Training Hub, users can easily fine-tune large language models using algorithms like LoRA, SFT, and GRPO without complex setup. πŸ› οΈ The article details how to run a LoRA fine-tuning job on Ray, focusing on SQL generation and model evaluation from a Jupyter notebook. A complete guide with examples is available in the Red Hat AI examples repository. πŸ”„ Ray offers...
Source: Red Hat Developer Blog
Fiona Waters

Improved failure reports on Red Hat OpenShift with the event-driven diagnostic operator

2026-08-21 13:18
πŸš€ Red Hat introduces an event-driven automatic first information report (FIR) collector designed to enhance failure reporting in OpenShift. This tool captures critical system context immediately during incidents, ensuring logs aren't lost or overwritten. It monitors key events, triggers data collection automatically, and stores logs off-site temporarily. This innovation helps teams analyze incidents effectively, improving response times and customer trust. Explore the project and its...
Source: Red Hat Developer Blog
Periyamaruthu Mohanraj

Red Hat Ansible development workspaces for governed automation content creation

2026-08-21 07:01
Red Hat introduces Ansible development workspaces, streamlining automation developer onboarding. πŸš€ By utilizing OpenShift Dev Spaces, developers can quickly access a fully configured VS Code environment in the cloud, enhancing standardization and security. Key benefits include faster onboarding, consistent tools, and centralized control over development environments. πŸ”’πŸ’» This approach eliminates common issues like version mismatches and local setup complexities. #Ansible #OpenShift #Automation...
Source: Red Hat Developer Blog
Leonardo Gallego