Docker Offload: Automate Workflows with Docker

Someone just dropped 𝗞𝘂𝗯𝗲𝗿𝗻𝗲𝘁𝗲𝘀 𝗳𝗼𝗿 𝗠𝗟 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝘀 on GitHub — and it is pure gold for anyone building AI infrastructure 🧠💻 It walks you through running, scaling, and deploying ML models on Kubernetes like a pro. Perfect for anyone tired of endless config files and deployment pain. This might be the new must-read repo for AI devs 👇 https://lnkd.in/g_WKtEk7

RAG, Memory & Context Engineering Clicked for Me: One of my biggest takeaways from the Google AI CTO bootcamp was this: LLMs don’t need more data. They need the right data, structured with intention. We explored the mechanics of RAG, memory design, and why agents “forget” even when you think they shouldn’t. The discussion about retrieval precision, summarization quality, and token-optimized context windows really resonated with me. The idea of treating the model’s “brain” like a carefully curated workspace with system instructions, user instructions, retrieved info, tool definitions, and conversation history changed how I think about building intelligent assistants. I now see context engineering as just as important as model tuning. #AICTO #AICloudTakeOff

🚀 97 % of devs still wrestle with PDFs like it’s 1999. IBM Research just changed the game: the Docling team (led by Peter W. J. Staar IBM Research Zürich) open-sourced a lightweight MCP server that turns any doc into an LLM-ready tool in 3 lines of code—no OCR hacks, no cloud lock-in. Agents can now “read” contracts, forms & reports on their own, 30× faster than OCR. Model Context Protocol (MCP) keeps doing awesome things; this time it’s an open-source doc parser that makes RAG pipelines actually reliable. Read how it works (and grab the repo) in the article below and on Towards AI—then go ship features while the rest are still copy-pasting. 👇 Link in first comment. https://lnkd.in/gn3sHmNz

2 Years of ML vs. 1 Month of Prompting” from the Levels.fyi blog. After two years of building custom machine-learning pipelines the article reports that a single month of sharply focused prompting with large-language models tied back to similar business outcomes. Prompt engineering shifts the execution model: less investment in labeled data and feature pipelines, more in prompt design, evaluation, and model alignment. Infrastructure, tooling and governance must adapt, a team previously optimized for MLOps now needs iteration frameworks, prompt libraries, and prompt-based observability. Adopting prompt-centric workflows does not eliminate complexity, it relocates it. Executives and platform teams must recognize the trade-off: shorter development cycles, but heightened demands on prompt lifecycle management, auditability and model behaviour oversight. More on the substack: 🔗https://lnkd.in/eqbVyPtf

🚨 New Episode: Building Platforms with Kaspar, this time with our friends at Cortex, Amazon Web Services (AWS), GitHub, Cursor, Datadog, Humanitec and more. I must say I underestimated how much effort it is to build and demo massive end to end IDPs but it's also totally worth it. In this episode I have a couple first times: I focus on Golden paths for Platform Engineers and I orchestrate VMs (!) which is something new after years of Kubernetes :); In this episode We explore: ✅ Detecting a policy violation (RDS overprovisioning) using Cortex ✅ Auto-remediating it in the background using Humanitec's Orchestrator ✅ Adding new capabilities to your platform (S3 support) without involving developers ✅ Using Humanitec Orchestrator with ECS runners to deploy infra ✅ Cortex scorecards, workflows, and natural language prompts (MCP server!) ✅ Claude Code + Cursor for AI-powered engineering This is the platform engineer’s perspective: how to codify golden paths, standardize infrastructure, and enable developer freedom without chaos. 🎥 Full episode in the comments 👀 Source code coming soon (this is actually exciting, we are packaging most of this as Terraform so you can replicate it! 💬 Questions, feedback, or your own platform story? Drop it in the comments. #platformengineering #internaldeveloperplatform #aws #terraform #humanitec #cortex #devex #idp #cloudnative #ai

This is the future of platform engineering - golden paths, AI copilots, and automated governance all working together. Kaspar Von Grünberg just showed what practical AI adoption looks like inside an IDP. Massive respect for how they’re showing, not just telling. #PlatformEngineering #EngineeringExcellence #AI

This is the future of platform engineering - golden paths, AI copilots, and automated governance all working together. Kaspar Von Grünberg just showed what practical AI adoption looks like inside an IDP. Massive respect for how they’re showing, not just telling. #PlatformEngineering #EngineeringExcellence #AI #Cortex

Missed GitOps Monthly? Catch the replay. 👀 In this episode, we dive into the Argo CD MCP Server—a powerful way to connect AI assistants like Cursor or Claude directly to your GitOps workflows. 🤖 Learn how to deploy, sync, and troubleshoot Kubernetes applications using natural language commands inside your IDE. 💬 Jiacheng Xu and Yiwei Gong walk through setup, security, and the future of AI-powered GitOps. 🚀 🔗 Watch on-demand: https://hubs.li/Q03PwcYR0 #GitOps #ArgoCD #AI #MCP #DevOps #Kubernetes #CloudNative

How can developers and AI collaborate to improve code quality? We’re building something special with our new MCP Server (and yes, it’s all about helping AI coding assistants improve code health). Join our upcoming webinar to empower yourself with tools for crafting healthy code, catching technical debt early, and keeping your workflow smooth. Here’s what you’ll see live: ✅ MCP Server in action. ✅ How to catch technical debt before it reaches production. ✅ Real-time feedback on code health in your IDE. ✅ Automated code health checks in pull/merge requests. ✅ How to ensure AI-generated code meets your standards. ✅ How to use an AI-assisted refactoring agent to iteratively improve code health. 🗓 Wednesday, November 26 🕝4–5 pm CET 💻 Online 👉 Sign up https://lnkd.in/gCPKKXjN #technnicaldebt #codehealth #codequality #aicoding

I’m excited to introduce an open-source AI system designed for intelligent document automation. 🧠 It brings together multiple specialized agents through a custom orchestration protocol, enabling document retrieval, summarization, classification, and automation within a single framework. Built with Python, FastAPI, and Haystack, it supports both cloud and local LLMs for flexibility and privacy. It’s equipped with a clean API, an intuitive Streamlit interface, and a robust RAG pipeline for accurate document understanding ! The source code is available on Github: https://lnkd.in/eVxNnR4T 🔗