Sign in

Samuel Karp

@samuelkarp.com
254 followers 147 following 44 posts

Anti fascist. Chair of the Open Container Initiative TOB. Staff Software Engineer @ Google. Opinions = mine.

PostsRepliesMedia
Samuel Karp @samuelkarp.com · 18/08/2026
oh I just got this too! (@tianon.bsky.social just linked me here when I sent this to him)
Screenshot of GitHub's issue UI. At the top it says "samuelkarp closed this as completed 1m ago". On the bottom, in the issue comment box, there is a callout that says "This issue was closed 57 years ago. Please consider opening a new issue or a new discussion instead of leaving a comment here."
221
Samuel Karp @samuelkarp.com · 02/02/2026
I wrote some basic integration with OneBusAway now.
A cash register display showing transit arrival data for two routes: the Swift Orange BRT line operated by Community Transit and the 1 Line light rail operated by Sound Transit.  Both are arriving within the next few minutes.
100
Samuel Karp @samuelkarp.com · 06/12/2025
I built myself a flight tracker that shows airplanes on an old cash register display! samuel.karp.dev/blog/2025/12...
Photograph of an LD220-HP display showing ADS-B data for SKW3853
414214
Samuel Karp @samuelkarp.com · 02/04/2025
This is pretty cool: a @containerd.dev shim to run LLMs with ollama. The shim augments the image by mounting in the ollama binary, generating a Modelfile as input to ollama (based on the model bundled in the image), and setting an entry point.
Mount Ollama binary

• Mount the Ollama binary from the host into the container

config. Mounts

append(config. Mounts, specs. Mount

Source:

"/usr/bin/ollama"

Destination:

"/usr/bin/ollama",

Type:

"bind".

Options:

[]string("rbind", "ro"},

})Mount OCI as a Modelfile

• Ollama needs to read the GGUF from a Modelfile file

modelPath: getTaskEnv ("OLLAMA_MODEL PATH", config.Process.Env)

modelfile: fmt.Sprintf("FROM %s", modelPath)

err = os.WriteFile(filepath.Join(tempDir, "Modelfile"), []byte (modelfile), 0644)
if err != nil {

return nil, err

}

config.Mounts append(config.Mounts, specs. Mount(

Source:

filepath.Join(tempDir, "Modelfile"),

Destination: "/tmp/Modelfile",

Type:

"bind",

Options: []string("rbind", "ro"},

})Serve the Model

Set the container args to serve the model using Ollama

config.Process.Args = [] string(

/bin/bash",

"-c".

strings.Join([]string{

"ollama serve &",

"while ollama ps> /dev/null 2>&1; do sleep 1; done",

"ollama create modelName + --file/tmp/Modelfile",

"tail -f /dev/null",

}, "\n"),
111
Samuel Karp @samuelkarp.com · 02/04/2025
This talk on LLMs is demonstrating why LLMs should be used with caution; KubeCon and DockerCon are not the same conference.
Slide with the following text:

Using Ollama

1. Download and install Ollama from https://ollama.com/download

2. Start Ollama server

ollama serve

3. Download your model

ollama pull llama3:8b

4. Use the model

ollama run llama3:8b

>>> What is the Kubecon?

KubeCon (formerly known as DockerCon) is an annual conference that focuses on cloud-native technologies, containerization, and Kubernetes...

>>>
67011
Samuel Karp @samuelkarp.com · 08/11/2024
We released containerd 2.0 this week, with new features, now-stable features, and deprecations. Join me next Friday at KubeCon to learn how to prepare for 2.0 in your production environments. We'll cover what's new and discuss how to detect/remediate any impact of removed features. sched.co/1i7qL
Promotional card for "What containerd 2.0 means for you" at KubeCon NA 2024 on November 15, 2024.
0277