ZenML MCP server now supports the full control plane. We added 12 new read-only tools covering entities AI assistants actually need to understand your ML infrastructure: → Deployments: Discover what's running, check status, pull bounded logs

Something I really miss in Raycast that was great in Alfred is the ability to trigger the command bar, then start typing and then immediately have the folders show up as you type.

I could move faster on side projects if I stuck with tools I already know. Lately I've been doing the opposite.

Every FastAPI server gets beautiful documentation for free. MCP servers get... nothing.

Building a "simple" abstraction is never simple. Schedules are a good example. At first glance: cron expressions, start times, done. Here's how ZenML's approach to schedules evolved:

I've been tracking my reading since 2012. Over 1000 books. This evening I wondered: how often am I picking books I end up satisfied with? Built a quick tool to find out—swipe right if satisfied, left if not. Rated all 1000+ books in one sitting.

I ran across a story from LinkedIn's engineering team recently that seems to illustrate the kinds of problems platform teams face when monitoring scheduled pipelines.

A couple of days ago I asked what local LLMs people were using for agentic work. I've now done the research and talked to people I know who run these models daily. Here's what I found.

Stale models hurt users. Twitter proved it. In 2018, Twitter's Cortex team built ML Workflows on Airflow to replace the ad-hoc scripts teams were using to retrain models. Before: "manually triggering and waiting for a series of jobs to complete." After: scheduled retraining that ran reliably.

It's been a while since I was last using local models in an agentic harness (think Toad / OpenCode etc). I want it more for documentation tasks + other local filesystem work rather than for coding per se.

I built a physical e-ink dashboard for monitoring ZenML pipelines. TRMNL is a small e-ink display—pulls data via webhooks and shows it on an always-on screen. Perfect for keeping an eye on ML pipelines without constantly checking the dashboard. github.com/zenml-io/tr...

I've been using Claude Code's slash command system to build little productivity tools, and yesterday I shipped one that scratches a long-standing itch: Detour. The problem: I'm deep in implementation work, and a tangential question pops up—"wait, how does this other module handle auth?"

I find myself creating these little scenario simulators a lot these days to explore little moments of curiosity. Today I almost slipped over on the way to the gym and so was wondering what makes certain shoes more grippy than others, and what exactly was going on with the physics of grip on ice.

Sentimental Value is such a great film. Going to see it for the third time next week. (Thanks @zoerosebryant for all the encouragement to give it a go over the holiday break!)

I posted a feature request in the @RepoPrompt Discord yesterday afternoon. By evening, @pvncher had shipped it.

I wrote previously about how Claude Canvas from David Siegel inspired me a bit to consider maybe an interface for ZenML pipelines, so I worked a bit in the background and now have a little interface for this to show off :)

I've been reading through recent entries in ZenML's LLMOps database. Three case studies show just how wide the spectrum is for how teams are actually running LLMs in production.

I've been paying attention to what people are building around Claude Code, and the interface experimentation has my attention. Case in point: claude-canvas by David Siegel. It uses tmux to spawn split panes alongside your CC session—giving the AI its own display surface for interactive TUIs.

ZenML Wrapped is out—a year-in-review for your ML pipelines. If you're running ZenML (especially with a team), this gives you a quick way to see what actually happened across all your projects last year. The setup: run a Python script → get a JSON file → serve a local site with your stats.

Something I found interesting about reading this thread is that it made me wonder how many people think of ChatGPT as primarily a tool for writing things for you I use ChatGPT/Claude/Gemini possibly 100+ times a day and NONE of those uses are to produce writing that I then share with other people

🔍 Seeking DeepSeek Prompting Insights for ML Labeling I'm gathering effective prompting strategies for DeepSeek models (R1/V3) specifically for large-scale data labeling operations.

Have been participating in a group that is doing fastAI part 1 for the first time. Took the opportunity in lesson 1 to use the text dataloaders / trainers to do a little miniproject to see how well fastAI does at classifying between the outputs of different LLM families.

Wrote up a few initial thoughts on the new iteration of Google's Deep Research mlops.systems/posts/2025-0...

Been meaning to use the latest Gemini(2.x series) models for a while now. Perfect timing by @strickvl.bsky.social with this set of practical examples in this standalone site. The only thing that could have made this even better for me would be direct curl examples, but maybe I'm asking too much. 😅

Little weekend project: 'Gemini By Example' – practical examples for building with Google's AI models 🚀 I built this simple site to complement Google DeepMind's Gemini SDK with focused, practical code examples - inspired by the "Go By Example" approach. geminibyexample.com

50,000

Just published a blog post about our new ZenML MCP Server - a tool that lets you chat with your ML pipelines using natural language! www.zenml.io/blog/chat-wi...

So excited to read this (just arrived). Had been waiting a long time.

In Berlin for the Amazon Web Services (AWS) GenAI Loft Week they're putting on. I'll be speaking about the lessons learned from the ZenML LLMOps Database for what it means to put LLMs in production. Do lmk if you're in town! Free registration for the Monday event: aws.amazon.com/startups/eve...

Working with historical documents in Dari, Pashto, and Arabic made me realise how challenging document translation remains, even with modern LLMs. The models often refuse to translate content (esp for sensitive topics), struggle with long documents and frequently lose context across page boundaries.

These are the kinds of NLP stories I love! TFW you "accidentally" train a great 2 MB (!) task-specific model 🤯 @strickvl.bsky.social

let's do this!

FYI...my understanding is that most, if not all of the dead have been buried in Lebanon, with families being told that repatriation of the remains could cost up to $3,000 US, which few of these families could ever afford. Governments are not offering assistance of any kind.

These deaths have been successfully buried. African leaders are not publicly commenting, fearing backlash at home for their failure to rescue citizens & to avoid ruining valued relations with Israel. Beirut based African diplomats (from at least two countries) have been instructed to stay mum.

Frustrated rant thread: Israel's 2023-2024 bombing campaign of Lebanon killed >4,000 people, the grand majority, civilians with no affiliation with any armed entity. Among them, anywhere from a dozen to 30 African migrants (maybe more), most of them Ethiopian women employed as domestic workers.

yeah this is really clear

Very nice recipe for creating a custom classifier on top of ModernBert. DeepSeek is used to generate labels to train on.

Holy moly. I'm trying to write an academic paper, and nearly every application I'm using is not only offering Generative AI as an option for writing, but *pushing it* -- pervading the design to the point where a simple misclick would make my content AI-generated. Here's why that's a problem. 🧵

A little over two months since my latest book came out! Society Girl — A Tale of Sex, Lies, and Scandal, with Tooba Masood Khan; a true crime story wrapped in a social history of Pakistan in the 70s. Available now in India and Pakistan, and on the Kindle store societygirlbook.com

📚 Notes on Ch.7 of @chiphuyen.bsky.social's AI Engineering: Fine-tuning isn't your first move - start with prompts & RAG. Memory matters, PEFT helps, and model merging opens interesting doors for multitask learning. Finetune for form, use RAG for facts. mlops.systems/posts/2025-0...

Realise that this stuff is a bit late 2000s web 2.0 style visualization, but have been enjoying checking in with trains.fyi every so often. A live / real-time map of passenger train locations, speeds etc in North America.

🎯 For the past few weeks, I've been sharing deep dives into our LLMOps database - exploring everything from RAG implementations to security practices. Today's post brings it all together AND introduces a new way to browse our 457 case studies! www.zenml.io/blog/llmops-...

Created a little tampermonkey script to handle YouTube video summarisation, since this is apparently what I now do on Sunday evenings :) gist.github.com/strickvl/ac2...

📚 Just kicked off our study group's read through of @chiphuyen.bsky.social's new AI Engineering. Chapter 1 unpacks how we're moving from traditional ML Engineering to AI Engineering - less about building models from scratch, more about adapting them effectively. mlops.systems/posts/2025-0...

📚 Just finished reading "Prompt Engineering for LLMs" by @jnbrymn.bsky.social & Ziegler. The final chapters brought everything together! I dropped my full notes in a blogpost but a higher-level summary follows. mlops.systems/posts/2025-0...

When using @anthropic.com's Claude.ai (i.e. web interface), do we know how longer texts (that appear as little 'document' icons beneath the chat interface when you paste something long e.g.) get inserted into the prompt? i.e. are they after the text in the box, or before it or what?

Hey, so not to be a downer, but if you see this go check your carbon monoxide detector. If you don’t have one, get one - they’re cheap and easy to install. My husband just attended the funeral of his soccer buddy and his whole family who were wiped out by CO on Christmas Day and it’s so damn sad.

Thank you to the U.S. Air Force for this document about its social media surveillance practices