About
Jamie Duncan is a recovering history major who has been working with and on Kubernetes…
Articles by Jamie
-
Why I Use Claude Code (Even Though I Work at Google Cloud)
Why I Use Claude Code (Even Though I Work at Google Cloud)
I work at Google Cloud. I use Claude Code for most of my development work.
5 Comments -
Building a Multi-LLM Publishing Pipeline: When One AI Isn't Enough
Building a Multi-LLM Publishing Pipeline: When One AI Isn't Enough
Today, I built something that probably sounds overcomplicated: a publishing system that uses two different LLMs to…
1 Comment -
The Craftsman's Paradox: Why AI Coding Agents Are Our Industry's Industrial Revolution
The Craftsman's Paradox: Why AI Coding Agents Are Our Industry's Industrial Revolution
I have a small workshop in my backyard. It was a COVID project, and is pretty well appointed, if not large.
8 Comments -
MCP and Local Skills/Extensions - Initial Thoughts
MCP and Local Skills/Extensions - Initial Thoughts
The AI tooling ecosystem has reached an inflection point. In 2025, we're witnessing a "AI Cambrian Explosion" with…
-
Building a SaaS Company as a One-Person Team: How AI Agents Became My Co-Founders
Building a SaaS Company as a One-Person Team: How AI Agents Became My Co-Founders
The Diet Pepsi has gone flat. It’s 2 AM.
6 Comments -
Dropbox Moving Data IN-House shows everyone it's about the best solution, not the newest buzzword
Dropbox Moving Data IN-House shows everyone it's about the best solution, not the newest buzzword
Today Dropbox released a blog post announcing their multi-year effort to bring their file storage in-house and almost…
1 Comment
Activity
5K followers
-
Jamie D. shared thisEvery few months a Chinese lab ships a model that matches a US frontier model for a fraction of the training cost, and the discourse treats it like a betrayal. DeepSeek did it. Then Qwen. Then Kimi. So I went and read what the CEOs actually said, not what got said about them. Sam Altman called DeepSeek R1 "invigorating." Dario Amodei corrected the record on what it actually matched, then argued for tighter export controls, not panic. Demis Hassabis said Google has "all the ingredients" to stay ahead. None of that reads like alarm. The people closest to the technology were the calmest people in the whole story. Ben Thompson called the real shift back in January 2025: once models are commodities, the advantage moves to cost structure, not the model itself. That's bad news for the compute layer and good news for everyone building on top of it. LLMs were never magic. Interacting with them well is a UX problem. Getting quality output out of them is a hygiene and testing rigor problem. They're just sharper tools than what we had before. Here's the case for treating them that way. https://lnkd.in/eXRnRS8E
-
Jamie D. shared thisGood idea and domain knowledge are never going to go away.Jamie D. shared thisThis is the brutal lesson a lot of AI builders are learning the hard way. AI didn’t just make development easier. It made the barrier to entry zero. And when anyone can ship functional apps in hours, technical execution stops being a moat. Building is no longer an advantage. The only edge left in the future will be sales and marketing. P.S. check out The Founder's Playbook for Winning in the AI Age 📚: https://lnkd.in/dtwjqH4W
-
Jamie D. shared thisIn The Triangle with the cool kids? Come hang out and let's talk about #Antigravity at DB in Cary!Jamie D. shared thisWe continue #AIEngineering season of #TriangleTechTalks with talk on Agentic Development with Google Antigravity by Jamie D. from #GoogleCloud Join us on July 22 at 5:30 at Deutsche Bank office in Cary, NC - use the link in the comments to register.
-
Jamie D. reposted thisThe Google Antigravity SDK?? 🤔 It gives you the same tools, agent loop, and context management that power Google Antigravity itself. But, what can you use it for? Check out Alexander Amin's video for a great intro to the SDK, as he hooks multiple agents into a Google Docs 📄 extension to help simplify his process! Great example, with source code too! https://lnkd.in/ePyiyYADJamie D. reposted thisLink in comments below :) Earlier this week, I broke down four ways to build AI agents. Now I'm going to show you how to implement /goal in your own Agent using the Antigravity SDK in Python. With very little code of your own! To make it a real use-case, I'm building an autonomous Google Docs agent I teased at the end of my last video. To keep it reliable, the architecture splits roles between a main agent, a researcher, and an auditor. Plus, it also shows how important and simple Human-in-the-loop can be built in the Antigravity SDK I also learned that LinkedIn really does not link if you share links in posts directly, so you'll find the link in the comments :) #AgenticArchitect #GoogleAntigravity
-
Jamie D. posted thisLLMs are the least efficient today they will ever be. Same statement will be true tomorrow.
-
Jamie D. shared thisWhen your knowledge base is simple, version-controlled Markdown, your agent reads documentation the exact same way it reads source code. No separate database, no extraction pipeline. When that knowledge base is a first-class citizen in your source code, you get better code faster from your agentic coding sessions. https://lnkd.in/esjQvw-MThe Machine is Your Reader: How to Build an LLM Wiki with AntigravityThe Machine is Your Reader: How to Build an LLM Wiki with Antigravity
-
Jamie D. posted thisHow I know I'll never visit Mars or the Moon - We can't get video conferencing on this planet right. How I know software developers aren't going anywhere - The last hour I just spent with Claude.
-
Jamie D. posted thisNvidia - Tokenmaxxing is the greatest! Nope. But they made a ton of money. Anthropic - GenAI will eliminate software developer jobs! Nope. But they made a ton of money and set up their IPO. Listening to the wolves when they tell you to leave the barn door open is always a bad idea. Calm down. The current GenAI technology is a _good tool_. It's not a panacea. And nothing will ever replace domain knowledge. It will never have a new idea in its current form.
-
Jamie D. shared thisI stopped feeding my BrandCast Labs marketing agent through a RAG store and turned the whole repo into a wiki the model reads directly. No retrieval step, just structured files. For a coding-adjacent agent that already reasons over a repo, it removed a whole layer of friction. It's turning out well. But building a knowledge base whose primary reader is a machine changes the rules. A human skims, tolerates clutter, and ignores the stale paragraph at the bottom of the page. A model does none of that. It reads everything you give it with equal weight. A contradiction you'd have glossed over becomes a coin flip in the answer. So the whole thing has to be built to stay true over time, on purpose, with tools that enforce it. A few rules that carry most of the weight: • Split knowledge (organized by subject) from content (organized by audience). Don't let your reference docs rot into half-finished drafts. • Write only what's true now. For a machine reader, the shelved "for context" paragraph isn't context. It's noise that competes with the truth. Delete it. • One home per fact. A parallel store always rots into a second, quietly conflicting source of truth. • Let tools do the enforcing. Discipline doesn't scale; linters do. The constraint turns out to be clarifying. Here's the full pattern. https://lnkd.in/e3_t4F4w
-
Jamie D. liked thisJamie D. liked thisFrom @gmail.com to @google.com🥳 I’m thrilled to share some exciting news-I have joined Google as an AI Customer Engineer. I’m looking forward to diving into this new journey, collaborating with brilliant minds, and working on technology that solves real-world challenges at scale. A big thank you to Kabir Ahuja and Veronica Gaddi for making the entire hiring process incredibly smooth, supportive, and seamless. Cheers to new beginnings! #Google
-
Jamie D. liked thisJamie D. liked thisBe the senior dev your past self needed. Comment your code. 📝
-
Jamie D. liked thisJamie D. liked thisLast Wednesday, we hosted more than 60 technologists at AWS Dallas to explore AI-driven team development workflows. We had attendees from aviation, advertising and marketing, financial, retail, and consumer goods, and I had the pleasure of moderating a discussion from a panel of industry leaders including Chaitanya Kommidi (American Airlines), Jyoti Rai (Southwest Airlines), Rahul Bellary (Epsilon), Bryan Tharpe from (Slalom), and Phil Austin (Jellyfish). Keynote speaker Paul Vincent walked us through AI-DLC, a set of open-source adaptive workflow steering rules for AI coding agents (https://lnkd.in/gUw35J9i), now in 2.0 Preview. These AI instructions allow enterprises to build with incredible velocity while maintaining the requirements, design, and decision traceability needed for large-scale systems engineering. Attendees had an opportunity to try it out using agentic assistants like Kiro and Claude Code. Special thanks to all our panelists and to co-sponsors Jellyfish and Slalom for helping to make this happen! For those who missed it, we're doing two more on July 22 in Chicago (https://lnkd.in/g-DarFMf) and Boston (https://lnkd.in/gPa3nn28).
-
Jamie D. liked this5 years at Google 🎊 1826 days of challenges, 1826 days of fun, 1826 days of friends, 1826 days of non-stop learning, 1826 days of passion, 1826 days of re-inventing myself, 1826 days of team collaboration, 1826 days of hard work, 1826 days of amazing people, 1826 days of tech, 1826 days of kindness, 1826 of empathy, 1826 days of joy, 1826 days of solving complex problems, 1826 days of a beautiful culture, 1826 of thought leadership, 1826 of working with great customers, 1826 days of gratitude, 1826 days of growth, 1826 days of the best place to work, 1826 days, and it's still day 1 for me... All of that and much more, is what working at Google feels like... Thank You #Google ! 🙌 🥲 🫡 🫶 🙏 🎉 🥳 🥂 #THANKFUL #GoogleCloud #Proud #Googler #Googleversary #5years #HalfADecade #Milestone #Achievement #Celebration
-
Jamie D. liked thisJamie D. liked thisHappy to share that I'm starting a new role as a Site Reliability Engineer at Google in Warsaw, Poland!
-
Jamie D. liked thisJamie D. liked thisI'm thrilled to share that I've joined Google's Security Engineering team to help secure the future of information discovery. As AI and agentic tools reshape how people discover information online and accelerate experimentation, it'll be all the more important to ensure that the data people entrust to those systems is handled safely, securely, and privately. And as the attack surface grows, so will efforts by determined adversaries to steal, hack, or misuse those tools. I'm incredibly excited to get to work countering them.
-
Jamie D. liked thisJamie D. liked thisGreat to see llm-d with vLLM performance featured in SemiAnalysis InferenceX, which was made possible by some recent work to decouple the llm-d router from the Kubernetes gateway by Google Cloud and support for slurm by IBM. You can learn more in Ezra Silvera's excellent post: "No Kubernetes? No Problem: llm-d Now Runs Anywhere": https://lnkd.in/gWV_Z3Sh cc: Abdullah Gharaibeh, Ezra Silvera, Akshay Ram, Fisayo Feyisetan (Okikiolu), Nathan Beach, Drew Bradstock, Carlos Costa, Pete Cheslock, Cameron Quilici, Robert Shaw
Experience
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Education
-
J. Sargent Reynolds Comm. College
-
-
currently working towards Associates in IT.
-
-
-
-
-
-
-
-
- Present
Licenses & Certifications
-
-
Foundations of Professional Identity
Coursera Course Certificates
-
-
-
-
-
-
-
-
CSM
ScrumMaster Alliance
Issued ExpiresCredential ID 108572
Volunteer Experience
-
Founder
Richmond, VA Linux Users Group
- 7 years 1 month
Education
Started the Richmond, VA Linux Users Group to help Linux enthusiasts and professionals get together, learn about new FOSS ideas and projects and learn from one another.
-
Director
Hack.RVA - Richmond, VA Hackerspace
- 1 year 6 months
Science and Technology
Spear-headed a reorganization effort within the group to modernize, streamline and grow the membership.
Publications
-
Protecting your Kubernetes deployments with Policy Controller
Google Cloud Blog
In November, the Kubernetes project disclosed a vulnerability which every Kuberenetes administrator or adopter should be aware of. The vulnerability, known as CVE-2020-8554, stems from default permissions allowing users to create objects that could act as a “Man in the Middle” and therefore potentially intercept sensitive data. If you are using a Google Cloud managed solution like Anthos or Kubernetes Engine (GKE), you can easily and effectively mitigate this vulnerability. In this blog, we’ll…
In November, the Kubernetes project disclosed a vulnerability which every Kuberenetes administrator or adopter should be aware of. The vulnerability, known as CVE-2020-8554, stems from default permissions allowing users to create objects that could act as a “Man in the Middle” and therefore potentially intercept sensitive data. If you are using a Google Cloud managed solution like Anthos or Kubernetes Engine (GKE), you can easily and effectively mitigate this vulnerability. In this blog, we’ll show you how.
Other authors -
-
OpenShift Storage for the masses
BrighTtalk
See publicationContainers are a developer's best friend. Quickly developing your code, dropping it into a container and running it anywhere you like can be intoxicating. But what happens to that container once it leaves your laptop? What do your Ops teams have to know, and what do they have to do to keep your containers happy and healthy and scaled up for production?
Join this webinar about how container-native storage offers developers the ability to dynamically provision storage for their modern…Containers are a developer's best friend. Quickly developing your code, dropping it into a container and running it anywhere you like can be intoxicating. But what happens to that container once it leaves your laptop? What do your Ops teams have to know, and what do they have to do to keep your containers happy and healthy and scaled up for production?
Join this webinar about how container-native storage offers developers the ability to dynamically provision storage for their modern applications. Exploring your containers' journey using practical examples, we'll cover:
-What actually happens when you create a Linux® container with Red Hat® OpenShift Container Platform.
-How container-native storage lets you dynamically provision storage for your modern applications.
-How to persist your Red Hat Gluster Storage container registry with Red Hat OpenShift Container Platform 3.5. -
OpenShift in Action
Manning Publications
Enterprise infrastructure has a lot of moving parts, and it's your job to keep everything running smoothly. OpenShift, an innovative enterprise infrastructure management system from RedHat, radically simplifies the day-to-day operation of deploying and maintaining large-scale applications. The OpenShift container management platform (CMP) uses Docker, Kubernetes, and other container-oriented technologies to ease cluster management, scaling, and upgrades. By providing an intuitive shell over…
Enterprise infrastructure has a lot of moving parts, and it's your job to keep everything running smoothly. OpenShift, an innovative enterprise infrastructure management system from RedHat, radically simplifies the day-to-day operation of deploying and maintaining large-scale applications. The OpenShift container management platform (CMP) uses Docker, Kubernetes, and other container-oriented technologies to ease cluster management, scaling, and upgrades. By providing an intuitive shell over established technologies, OpenShift gives you the reliability and stability you need with the ease of use you want.
Other authors -
-
How to Rapidly Enable Digital Transformation in your Organization
Red Hat, Inc.
See publicationContainers are fueling Digital Transformation at a global scale by delivering technology to users more rapidly than any other solution at our disposal.
-
Enabling Open Source High Performance Workloads with Red Hat
HPCWire
See publicationUsing Red Hat Technologies to innovate in the Scientific Computing world
-
Pushing the Boundaries: High Performance Computing in Government
FedInsider
See publicationHow high performance computing and cloud are pushing the boundaries of science, discovery and technology in government today.
-
Atomic @ Scale
Southeast LinuxFest
See publication2016 SouthEast LinuxFest
Jamie Duncan
Atomic @ Scale
Projects
-
The K Files Podcast
-
See projectEach week we take on a different topic in the Kubernetes community and the IT industry. Depending on the week we may find ourselves knee-deep in kernel headers, magic quadrants, or startup pitch decks. Our goal is to close each file with a sense of completeness and satisfaction for us and for anyone listening.
-
SOSCleaner
-
See projectSOSCleaner helps users in environments that have data export restrictions clean up an sosreport so it can be safely uploaded to a support group for analysis. It is currently under consideration for inclusion in Red Hat Enterprise Linux.
-
NewRHELic
-
See projectA Red Hat Enterprise Linux Plugin for New Relic.
This provides a more granular look at Red Hat Enterprise Linux server performance data, and should be compatible with its downstream derived projects (although not always tested).
Honors & Awards
-
Chairman's Award
Red Hat, Inc.
This annual award program recognizes associates with contributions so strong, so superior, and worthy of merit and acknowledgment that they have been nominated by their peers to receive company-wide recognition for reasons like:
Living the Red Hat values
Creating a positive customer experience
Being a leader
Promoting collaboration
Demonstrating functional/technical expertise
Languages
-
PHP
Professional working proficiency
-
Python
Full professional proficiency
-
Perl
Professional working proficiency
Recommendations received
1 person has recommended Jamie
Join now to viewOther similar profiles
Explore more posts
-
Tim Bertrand
Multi-cloud scale needs an entirely different approach to connectivity. Relying on sidecar-heavy setups to bridge Kubernetes deployments and legacy data centers introduces unnecessary latency and management overhead. The "mesh tax" is real, but it is no longer a requirement for enterprise security and routing. True scale requires centralized orchestration. Universal Mesh replaces complex sidecar fleets with strategic Edge Gateways. It provides a single, unified layer to seamlessly connect modern and legacy environments without the performance penalty. Jakub Suchy and Ron Northcutt are hosting a live session to demonstrate this architecture in action. Join them as they cover how to dynamically route around overlapping IP conflicts, unify your infrastructure, and eliminate the mesh tax from your operations. Sign up here: https://lnkd.in/e2uRT5Rt HAProxy Technologies
-
Frank Contrepois
Yesterday's hyperscaler news that caught my eye is an #AWS one It has some hints of #FinOps too. EC2 Auto Scaling announces warm pool support for Auto Scaling groups that have mixed instances policies Starting today, you can add warm pools to Auto Scaling groups (ASGs) that have mixed instances policies. With warm pools, customers can improve the elasticity of their applications by creating a pool of pre-initialized EC2 instances that are ready to quickly serve application traffic. By combining warm pools with instance type flexibility, an ASG can rapidly scale out to its maximum size at any time, deploying applications across multiple instance types to enhance availability. https://lnkd.in/evWzkSfS
-
dstack
Since launching fleets last year, they’ve become one of the most impactful ways to simplify GPU provisioning across cloud and on-prem. Over the past year, we've added: * Native Kubernetes support * Elastic fleets that auto-scale as team needs change Read more about how fleets work: https://lnkd.in/ehz-77hf
-
Ted Jones
📈 Whether you're a system administrator, developer or SRE, staying ahead in cloud-native computing means embracing the tools that give you a competitive edge. One of those tools? eBPF (Extended Berkeley Packet Filter) 🎉 From enhanced security and networking to next-level observability, eBPF is transforming how modern platforms monitor and understand operating systems and applications. That’s why SUSE integrates eBPF into its observability stack—delivering real-time insights, performance visibility, and operational control across environments. 📖 Find out more: https://okt.to/RQPod5 #eBPF #Observability #CloudNative #SUSE #DevOps #SRE #Linux #InfrastructureMonitoring #PlatformEngineering
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content