Daniel Liden

@danliden.com

Developer Advocate @Databricks

We—especially the left—lack much capacity for utopian imagination. Not "here's the list of bullet points we're aiming to achieve." But the ability to imagine a plurality of complicated futures where technology ends up being a net good. That outcome seems impossible without healthy imagination.

It has been a lot of fun working with Omnigent in the weeks since its release. Policies were the component I had the least reason to use in my daily coding use, but I still wanted to understand them. I tested & wrote up a few different ways to use policies here: www.danliden.com/posts/202607...

Allow, ask, deny: contextual policies in Omnigent

An introduction to Omnigent policies: contextual ALLOW/ASK/DENY guardrails for AI agents, with built-in, custom Python, and CEL examples.

danliden.com

The goal with my rlhf book is to make the "home on the internet" for the next generation learning post-training. That's why I'm doing all formats (lectures, code, book, discord, model completions... & ofc blog of interconnects). A hub is more lasting than non-fiction writing. Join! rlhfbook.com

Bild

blockclubchicago.org/2026/06/23/c... 1. Seems like a silly argument against bike lanes in the first place 2. As a biker, I oppose protected bike lanes. Conventional bike lanes make bikers more visible and give space to bike around obstacles. More ways to respond to situations = safer!

Chicago's Bike Lanes Don't Hurt Businesses, City Report Finds

As protected bike lanes face opposition in some neighborhoods, city-backed case studies found lanes in a half-dozen corridors have not contributed to a negative impact on economic activity.

blockclubchicago.org

Given today's LiteLLM supply chain attack, what are people's preferred development environment sandboxes on MacOS these days? I think it's time I started running my development environments somewhere where rogue code can't steal all my ~/... credential files

A reminder to the news media: “conflicting accounts” is what you say BEFORE the incontrovertible video evidence appears. After that, your job is to ask why one side is lying, not to repeat the lie and pretend no one knows the truth.

The last 6 months of 2025 were a defining era for MLflow! 🚀 ✅ 𝟭𝟰 𝗥𝗲𝗹𝗲𝗮𝘀𝗲𝘀: Focused on LLM Agentic observability, evaluation, & scalability ✅ 𝗡𝗲𝘄 𝗙𝗲𝗮𝘁𝘂𝗿𝗲𝘀: GenAI evaluation, unified tracing & more ✅ 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝘁𝘆 𝗚𝗿𝗼𝘄𝘁𝗵: 23 Ambassadors across 17 countries Dive in ➡️ www.linkedin.com/pulse/celebr...

Celebration of Innovation: A Retrospective on the MLflow Community's Last Six Months of 2025

Authors: Carly Akerly, Daniel Liden & Jules Damji MLflow 3.x and the GenAI Revolution The past six months have marked a transformative era for MLflow, with a total of 14 releases.

linkedin.com

Playing around with #emacs org-mode tables. I ultimately want an easy system where I can use capture templates to easily log new rows of data to tables in particular buffers/files.

Bild

Every so often I rediscover the immense usefulness of registering window configurations in #emacs . <C-x r w> to register the configuration, navigate away, return with <C-x r j>. Much easier than trying to reconstruct a complicated window arrangement after e.g. deleting windows.

MLflow has added an MCP server in the most recent release, MLflow 3.4.0. You can e.g. use natural language to find and diagnose failing agent traces w Clauded Desktop. This is a huge step toward making MLflow a tool usable by agents. Blog on setting this up coming soon. #mlflow #ai #agents

Bild

Seems like this is a common experience at this point—I ran into an issue where Claude Code was struggling, switched to Codex and was able to resolve it quickly. But the UX of codex is much worse in its present form. Still—really incredible to have have both these tools to work with right now!

🚀 Join the next MLflow Community Meetup on Oct 1 at 4PM PT! 🔹 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗦𝗺𝗮𝗿𝘁𝗲𝗿 𝗘𝘃𝗮𝗹𝘀 𝘄𝗶𝘁𝗵 𝗧𝗿𝗮𝗰𝗲-𝗔𝘄𝗮𝗿𝗲, 𝗙𝗲𝗲𝗱𝗯𝗮𝗰𝗸-𝗔𝗹𝗶𝗴𝗻𝗲𝗱 𝗝𝘂𝗱𝗴𝗲𝘀 🔹 𝗞𝗲𝗲𝗽𝗶𝗻𝗴 𝗘𝘃𝗮𝗹 𝗗𝗮𝘁𝗮𝘀𝗲𝘁𝘀 𝗥𝗲𝗹𝗲𝘃𝗮𝗻𝘁 𝗮𝘀 𝗬𝗼𝘂𝗿 𝗔𝗽𝗽 𝗖𝗵𝗮𝗻𝗴𝗲𝘀 Bring your questions about dataset management, evaluation workflows! ✅ RSVP 👉 luma.com/mlflow-1001 #opensource #oss

MLflow Community Meetup · Luma

Join us for the next MLflow Community Meetup on October 1 at 4PM PT! Ben Wilson, MLflow Maintainer, will dive deep into: Building Smarter Evals with…

luma.com

The first-ever Open Lakehouse + AI Meetup in Amsterdam was a huge success! 🙌🇳🇱 The evening brought together the open source and data engineering community to explore the latest advancements in open lakehouse and AI architectures. 🚀

BildBildBildBild

📣 MLflow 3.3.0 is now available! This release introduces several major features and improvements: 🔹 𝗠𝗼𝗱𝗲𝗹 𝗥𝗲𝗴𝗶𝘀𝘁𝗿𝘆 𝗪𝗲𝗯𝗵𝗼𝗼𝗸𝘀 🔹 𝗔𝗴𝗻𝗼 𝗧𝗿𝗮𝗰𝗶𝗻𝗴 𝗜𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻 🔹 𝗚𝗲𝗻𝗔𝗜 𝗘𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻 𝗶𝗻 𝗢𝗦𝗦 🔹 𝗥𝗲𝘃𝗮𝗺𝗽𝗲𝗱 𝗧𝗿𝗮𝗰𝗲 𝗧𝗮𝗯𝗹𝗲 𝗩𝗶𝗲𝘄 🔹 𝗙𝗮𝘀𝘁𝗔𝗣𝗜 + 𝗨𝘃𝗶𝗰𝗼𝗿𝗻 𝗦𝗲𝗿𝘃𝗲𝗿 🔗 Check out the release notes: github.com/mlflow/mlflo... #oss

📣 MLflow Office Hours — Wednesday, Sept 10 Connect directly with #MLflow maintainers and contributors for live Q&A! Bring your production challenges or your latest #LLM and #GenAI experiments—this session is dedicated to hands-on technical discussion and feedback. Save your spot ➡️ lu.ma/mlflow-910

MLflow Office Hours | September 10 · Zoom · Luma

Join us for the next MLflow Office Hours on Wednesday, September 10, for an open Q&A session with MLflow maintainers and contributors! 🎊 Whether you're…

lu.ma

Anyone regularly use #emacs ediff? I just used it to work through a (simple) merge conflict and it was quick and easy, but I've never really invested the time to seriously learn it. Worthwhile?

Join us for the next MLflow Community Meetup! 🥳 This month, we’re excited to bring you the latest updates and insights from the MLflow ecosystem, featuring: ​🔔 Webhooks in MLflow 🧪 Managed MLflow Features Now in OSS 📅 August 20 @ 4:00 PM PT 📍 Live on X, YouTube, & LinkedIn 🔗 RSVP: lu.ma/mlflow820

Bild