William (Bill) Kennedy

Here are my 2 #golang talks from @GopherCon Understanding Escape Analysis to Speed Up Your Code www.youtube.com/watch?v=FUm0... Go’s Trace Tooling and Concurrency www.youtube.com/watch?v=Gqo0...

Updated the Weather app on ArdanLabs AI demo site. For sure check out the Connect 4 game. This is running in Cloud Run, using NVIDIA GPUs, and #Kronk to run model inference. #golang ardanlabs.ai

How can one AI API support different models and their unique requirements? 🤖 In this clip from @goinggo.net ’s Ultimate AI course, explore flexible model inputs, streaming response loops, and real-time tool calling, including a demo with a tiny 0.6B model. 🎥 Watch the clip: youtu.be/jrM2bUlRoys

A year ago I had finished work on a Connect 4 game where you played against an embedding and language model. I converted that app to use #Kronk running in Cloud Run with a NVIDIA GPU. Model inference integrated in a Go service running in the cloud! ardanlabs.ai

YouTube short of the week is here!💻 This week revolves around “Being Able to Tell Your Story” from @goinggo.net 's podcast episode with featured guest Ken Stott! 🤓 🎥: youtube.com/shorts/tckyP... #ArdanLabsPodcast #Shorts #YouTube #SoftwareEngineering

A model on integrated graphics, playing a full game in real time, while my own chat model crawls at a speed I don't want to name. Nothing leaves the machine. This is how local AI should be demoed, not with a 24GB card.

Austin, just 1 week until the next Cloud Native meetup! 🚀 Join @goinggo.net as he dives into open-source AI inference, model serving, and Kronk in his featured talk “Ultimate AI: Go with your own intelligence.” 📅 Oct 22 | 📍 Station Austin 🔗Register: ocgroups.dev/cncf/group/3...

And now... @goinggo.net at #golangsyd... On Kronk, Bringing @deadprogram.com's yzma together. #golang

Is a bigger AI context window always better? 🧠 In this clip from our Ultimate AI Course, @goinggo.net breaks down Dense, MoE, and Hybrid architectures, why models lose focus, and how to manage context when running open-source LLMs locally. 🎥 Watch the clip: youtu.be/TMarfNTvK14

How cool is this! #Kronk #golang Kronk Jev Game is an endless runner. Play the top half of the screen. A small AI model plays the same map in the bottom half, at the same time. No cloud. No API key. Nothing leaves the machine. The model runs on my laptop's integrated GPU. lnkd.in/p/ehuYbP7J

いい写真。 #tinygo_conf

New #Kronk release v1.32.9 #golang

🤖 Two new AI courses are now included in our Go bundles! → Ultimate AI by @goinggo.net → Practical Go Development with AI Agents by @tebeka.bsky.social Explore the bundles: www.ardanlabs.com/training-bun... Already a member? courses.ardanlabs.com/users/sign_in

A great evening with the Go community in San Francisco! 🐹🌉 @goinggo.net took the stage to talk about self-hosted LLMs in Go, exploring what’s possible when AI runs closer to the code, the hardware, and the engineer. Thanks to everyone who joined and GoSF for a great meetup! 🙌

San Francisco, we’re meeting tomorrow! 🦾 Join us at Antithesis for the next GoSF meetup: 🦾 @goinggo.net : Self-Hosted LLMs in Go ☕ Live Cup o’ Go with Shay Nehmad 📍 Oct 7th | 5:30–8 PM PDT & we’re raffling an Ardan Labs training bundle 👀 www.meetup.com/golangsf/eve...

gVisor is being donated to CNCF for a more open governance model! gvisor.dev/blog/2026/10...

Two NEW AI courses are now available ! 🤖 • Ultimate AI by @goinggo.net • Practical Go Development with AI Agents by @tebeka.bsky.social Learn more → www.ardanlabs.com/training-bun... Already a member? The new courses are ready for you courses.ardanlabs.com/users/sign_in

New #Kronk release v1.32.8 #golang

One week to go! 🦾 GoSF is back in San Francisco on October 7th, with @goinggo.net talking about Self-Hosted LLMs in Go. Go, AI, networking, & community. 📍 Antithesis 🕠 5:30 PM PDT 👉 www.meetup.com/golangsf/eve...

@squillace.bsky.social did you see my recent jeyzma.com yet?

Bystander Society by @aleintech.com alekennedy1.substack.com/p/bystander-...

Yesterday, @goinggo.net joined the GolangNYC Meetup in New York City. Great conversations, a room full of developers, and plenty of ideas shared around building with Go. Thanks to everyone who came out and made it a great meetup! 📸 Sharing a few moments from the event below.

Go + AI is coming to NYC! Join @goinggo.net at GolangNYC on Sept. 23rd for networking / a talk from his Ultimate AI curriculum to explore what building AI systems with Go can look like. 📍 New York City 📅 September 23rd | 6-8pm RSVP: www.meetup.com/golanguagene...

What if your Go application could handle AI inference directly? @goinggo.net walks through the #Kronk Model Server and how it handles: → Local & remote GPU inference → Real-time VRAM monitoring → In-Memory Caching → Hugging Face models → Speech + image generation 🎥: youtu.be/Sv0ZhQjaa1M

v1.32.7 Released Version 1.32.7 brings image generation into the model server with new API endpoints and BUI screens, a new OpenAI-compatible audio translation endpoint, Wan2.2 S2V audio-conditioned video support in the Malina SDK, and llama.cpp upgrades through b11017. github.com/ardanlabs/kr...

If you are in the SFO area on Wednesday Octber 7th I will be speaking at the Go meetup and talking about all things #Kronk and #golang. www.meetup.com/golangsf/eve...

Quick overview of the #Kronk model server.

Thank you very much to #cupOGo for having me as a guest last Friday to talk about @tinygo.org Check it out here: cupogo.dev/episodes/a-b... #golang #tinygo #embedded

Version 1.32.6 Released llama: b10896 whisper: v1.9.3 stablediff: master-849-d04e895 kronkai.com

YouTube short of the week is here!💻 This week revolves around “Choosing Priorities in a Career Path” from @goinggo.net t 's podcast episode with featured guest Jonathan Amsterdam! 🤓 🎥: youtube.com/shorts/IhkkO... #ArdanLabsPodcast #Shorts #YouTube #SoftwareEngineering

That’s a wrap on GopherCon LATAM! 🇧🇷 Thank you to everyone who joined us, stopped by, and shared conversations with the Ardan Labs team. A special thank you to the organizers, @goinggo.net , and the Ardan Labs engineers who attended and made the conference a great experience. Until next time! 👋

Check out the new #Yzma website and try Yzma in the browser while you are there! #golang yzma.ai

@goinggo.net is kicking off a workshop: Ultimate AI, at @gophercon.com LATAM today. 🇧🇷 A room full of engineers, Go + AI, and lots of hands-on learning. Here are a few snapshots from the workshop 👇 #GopherConLATAM #golang #ArdanLabs

YouTube short of the week is here!💻 This week revolves around “Discovering Go While Working at Google” from @goinggo.net's podcast episode with featured guest Jonathan Amsterdam! 🤓 🎥: youtube.com/shorts/FZLlb... #ArdanLabsPodcast #Shorts #YouTube #SoftwareEngineering

Checkout the current family of #Kronk stickers coming to a #golang event near you. Check out the travel schedule on the Kronk repo. github.com/ardanlabs/kr...

In my neverending attempts to make these models behave well enough for real usage, @dlsniper shared these non-thinking settings with me this morning and I have to say the model is usable today. That can change by the hour, but it's usable!

The AI takeover continues in Episode Two. In this scene special guest star @boursiquot.com asks Phineas a question with hilarious results. See the only show with robot actors run on @officialarduino.bsky.social UNQ Q boards & @golang.org / @tinygo.org code on @github.com youtu.be/Pny5TFP2r3U

#Kronk v1.32.1 Released Removes the adaptive MTP draft-count algorithm and its runtime reporting. MTP speculative decoding now always runs with the configured fixed `ndraft` value, while classic separate-draft speculation retains its acceptance-EMA adaptive throttling.

#Kronk v1.32.0 published Brings adaptive draft-count tuning for MTP speculative decoding, an upgrade to llama.cpp (yzma v1.25.0), new codegen-benchmark tooling, and fixes to GLM tool parsing and embedded MTP compatibility detection. github.com/ardanlabs/kr...

Latest installment of Talking Heads From The Year 2053 - Episode Two has been released by @deadprogram.com www.youtube.com/watch?v=Pny5...

I wrote a program that uses OpenCode to have different models write a Go based tic-tac-toe game. This take a wee bit of time to run for all models. The Gemma4 and the Qwen3.8 models always do well. I wish the Qwen3.8 ran faster. Tomorrow Qwen3.8 Flash is being released!

This is a simple example of why you need to be looking at diffs when the model/agent is making code changes. This is why vibe coding is going to cause all sorts of problems over time with codebases. There needs to be a human in the loop.

My new article, the Kronk SDK, is live. 3 important components: → prep: messages validated, templated and tokenized into a job → KV cache: restore the prefix, decode only what's new → engine: makes your tokens tick concurrently internals-for-interns.com/posts/kronk-... #golang #AI

Ok Kids, new release of Kronk coming soon to fix some performance issues. This model is outstanding and the best I've seen in a long time. It's a Qwen based model published about a month ago. It has been better than the Qwen3.8 27B for sure. huggingface.co/unsloth/Orni...

Get ready #golang v1.27.0 will be released today! groups.google.com/g/golang-ann...

v1.31.5 release of #Kronk now available. Best version yet. Plus updated website with new Bucky and Malina pages. kronkai.com

Latest version of #kronk released today gives you a view of how the slots are running and the IMC caches. Check it out!