InfoQ

Docker Cloud Sandboxes provide secure, hosted execution environments for running AI coding agents on Docker-managed infrastructure. Built on hardware-enforced microVM isolation, the platform provides a consistent execution environment and unified CLI workflows for seamlessly moving workloads from local machines to the cloud. 🔗 Learn more: https://bit.ly/46X127V #Docker #AIAgents #Cloud #DevOps #InfoQ

The spec says one thing - real-world implementations say another. Building the YK JIT compiler revealed a hard truth: Python, Ruby & Lua specs are inherently incomplete. The real source of truth? The standard C reference implementation. Laurence Tratt explains why C interpreters dictate real-world behavior. 🔗 Watch now [transcript included] - https://bit.ly/4z1Fbce #ProgrammingLanguages #Compilers #Python #Ruby #Lua #SoftwareEngineering #InfoQ

#DoorDash cut high-latency LLM calls by 90% and reduced verbal abuse incidents by 50% - all with one core architectural pattern. In this talk, Software Engineer Bruna Pereira breaks down how DoorDash built SafeChat and evolved it into a flexible, content-agnostic LLM orchestration platform. 🎬 Watch now: https://bit.ly/4xsQjOc #SoftwareArchitecture #SystemDesign #MachineLearning #LLM #InfoQ

What happens when software stops being deterministic? Explore the enterprise AI race - and the trade-offs around reliability, ethics, AI-generated code, open vs. proprietary models, and software supply-chain sovereignty. 🎧 Listen now: https://bit.ly/3VmGXp0 Panelists: Meryem Arik, Clara Higuera Cabañes, PhD, Jeff Smith & Olimpiu Pop #AI #AIAgents #OpenSource #SoftwareEngineering #InfoQ

Security controls in an agentic system can sit at several layers: the model boundary, the data flow, tool access, or the sandbox around execution. The architectural decision is how those controls work together, and what should happen when one fails. Those are the kinds of decisions inside the InfoQ Certified AI Security & Privacy Engineering Program. Across five weeks, participants trace sensitive-data flows, threat model and red team an AI workflow, work through guardrails and sandboxing, and test controls using observability and evaluations before moving into governance and ownership. The capstone is a security and privacy assessment for an AI product architecture, covering data exposure, threats, controls, testing, and ownership. The program is facilitated by Katharine Jarmul, author of Practical Data Privacy (O'Reilly), with around a decade in machine learning and AI and roughly eight years focused on privacy and security. The next cohort starts October 26. Review the program and syllabus: https://bit.ly/4iUDnw2 #AISecurity #DataPrivacy #ThreatModeling

FIVE WAYS TO USE AI CODING AGENTS TO IMPROVE YOUR SOFTWARE ARCHITECTURE Automated code generation is not a new concept, but AI Coding Agents are dramatically faster at coding than anything that has come before. This speed creates huge benefits, but it also creates unique problems because its very easy to lose control of the quality of the output. This loss of control is especially true of the architecture; if you only feed the AI with functional requirements, the AI isn't somehow going to make sure the architecture is sound. You have to feed it with specific architectural goals, such as measurable Quality Attribute Requirements (QARs) and trade-offs. The AI-generated code needs to be tested to evaluate QAR satisfaction. Using AI coding agents to develop resilient, scalable, secure systems is in its infancy. There are, however, some good ways to jump into it, to make the most of experiences without wasting too much time simply trying random things. Here are some suggestions we have found useful in starting the journey toward using AI coding agents to develop systems with a sound architecture: 1. Modern architectures often use legacy services for specific tasks, but these services may lack accurate documentation and using them can be risky if you dont understand them well. AI coding agents can help close that knowledge gap. 2. You can use an AI coding agent to find and suggest fixes for common architectural problems that are organization-specific or generic. 3. AI coding agents can be used to identify and patch security vulnerabilities, which is especially useful when the architecture includes open source packages 4. AI coding agents can free teams to experiment, but they need constraints to conform to specific, measurable architectural goals and trade-offs. They can then create prepackaged shell applications that form the foundation for developer-driven prototypes. 5. An AI coding agent provides an efficient, fast way to generate Minimum Viable Architectures (MVAs) and evaluate the MVA code through measurable tests. Interested in learning more about this? Read this InfoQ article authored by Kurt Bittner, Todd Miller and me, and reviewed by Daniel Bryant at https://lnkd.in/gfxgMHMc #SoftwareArchitecture #CodingAgents #AIEngineering

I had a fantastic time recently talking with Michael Stiefel on the InfoQ Architects Podcast about some of my favorite topics: DevEx, Engineering Culture, and Psychological Safety and how I think these topics are more important now than ever. I hope you'll check it out! https://lnkd.in/gJp64s93

99.99% of engineers aren’t building AI models. They’re building context! The future of DevOps isn't model training - it's testing and benchmarking agent context with the tools you already know. Stop testing models. Start testing context. 🗓️ Full presentation goes live on InfoQ September 30. #AIAgents #ContextAsCode #DevOps #InfoQ

#AICodingAgents can move fast. But… your architecture might not survive the ride! Speed without architectural guardrails can create new problems. Give agents the context they need to meet quality goals - and use them to explore, validate & evolve software architecture. In this #InfoQ article, Pierre Pureur, Kurt Bittner & Todd Miller share practical guidance for making AI Coding Agents work with architecture, not around it. 🔗 Read now: https://bit.ly/46M77Ef #AI #SoftwareArchitecture

An agent that generates the code should not be the only thing deciding whether that code is good enough to merge. Independent verification is one of the decisions inside the InfoQ Certified AI-Assisted Engineering Program. Over five weeks, participants build a harness against a real brownfield codebase. The work covers onboarding an agent with least-privilege permissions, adding sensors and characterization tests, separating generation from review, moving verification into CI, and measuring what changed. Participants leave with a harness they built and five weeks of their own logged results to compare against what they predicted at the start. The program is facilitated by Zichuan Xiong, who has led architecture and delivery work since 2008 and now builds agentic AI systems for software operations, and Premanand Chandrasekaran, who has spent two decades leading engineering teams with a focus on continuous delivery and internal quality. The next cohort starts October 19. Review the program and syllabus: https://bit.ly/4y1weOA #AIAssistedEngineering #SoftwareEngineering #ContinuousDelivery