Joe Schmid retweeted
💯 The next generation of BI tools needs to tackle both (1) scalable analysis of unstructured data, and (2) *efficient* communication of the analysis to *humans* to inform key business decisions. Come apply for a PhD with me if you want to build the foundations and systems for this new era!!
Joe Schmid retweeted
Every company will have a context layer.
Today, we're launching Prefect Horizon to make that possible.
The context layer is where AI agents interface with your business. It's where teams expose their proprietary data, tools, and workflows to autonomous systems. It’s where you define what agents can see and do, and where you curate and distribute the context that makes automation useful and safe.
We know what this requires because we created FastMCP, the standard framework for working with the Model Context Protocol. We've watched it grow to a million downloads a day.
But we've also seen the protocol's limits. MCP describes how to *build* a server, not how to govern it at scale. It doesn't tell you how to deploy it, secure it, or share it with 5,000 colleagues.
We built Prefect Horizon to turn MCP from a protocol into a platform:
🚀 Horizon Deploy: Managed MCP hosting that gets you from PR to governed endpoint in 60 seconds.
📚 Horizon Registry: A central catalog for all your MCP servers, including first-party, third-party, and our new remix servers that compose multiple sources together.
🔐 Horizon Gateway: The control plane enterprises need: role-based access control down to the tool level, audit logs, and usage visibility.
💬 Horizon Agents: The last mile. Give business users an agentic interface to your company without ever knowing what "MCP" is.
This is automation for the context era.
Welcome to Horizon.
Joe Schmid retweeted
Youngest grandson cheering on Team Canada tonight……taking a break from outdoor rink fun 🇨🇦🇨🇦🇨🇦🇨🇦
Joe Schmid retweeted
Thrilled to share four milestones we hit in July:
🚀 Record‑breaking quarter: we just posted the strongest in company history.
🏁 Profitable: we’ll affirm this milestone in the next few weeks.
🤩 Customer momentum: +70 % net growth in just six months.
📈 Exploding usage: Prefect and FastMCP are both seeing record growth.
None of this happens without an extraordinary team. I can’t wait to celebrate together at our annual gathering in two weeks, here in DC — then roll up our sleeves for @PrefectIO’s next chapter. Onward!
Joe Schmid retweeted
DocETL continues to evolve as a powerful system for LLM-powered data processing & analysis pipelines. Since I'm at UIST this week, this deep dive focuses on UI/UX insights for transforming unstructured data with LLMs--on complex tasks and the art of prompt engineering 🧠🔧
Joe Schmid retweeted
Ok time to come out of my serverless GPU closet:
@modal_labs is the best, and it’s not even close.
We had many competitors at Banana, with users notorious for always shopping around.
The one exception was Modal. A black hole of user retention in an otherwise commodity market.
Joe Schmid retweeted
We were able to embed all of Wikipedia in < 15 minutes on Modal Labs for ~ 17$.
Here's what this actually means for an organization:
1. Unrestricted Rates: Eliminates the bottleneck of rate limits for large-scale operations.
2. Rapid Experimentation: Allows quick iterations of data chunking experiments, reducing wait times from hours to minutes.
3. Fine-tuning: Use of the same hf infrastructure for model fine-tuning, speeding up experiment cycles, and deployment.
Reducing experiment timelines from hours to approximately to minutes is something that has compounding effects!
take a look! modal.com/blog/embedding-wik…
Joe Schmid retweeted
Building the product you wish you had isn’t enough for a successful startup.
Other people have to wish for it, too.
Joe Schmid retweeted
Big Data Workflow Automation with @PrefectIO and Coiled🔥
In this post, we discuss when and how to use Dask and Coiled to leverage parallelism in your Prefect workflows, either locally or in the cloud.
coiled.io/blog/big-data-work…
Joe Schmid retweeted
Dask gets deployed in many ways.
💻Some folks run on a single machine.
⌨️Some SSH to more machines.
🏢Some run on HPC.
☁️Some scale to the cloud.
🚀Some build entire Dask platforms.
Thought I'd write up a common path that I've seen many folks take.
blog.dask.org/2021/06/01/das…
Joe Schmid retweeted
Data and AI operations startup Coiled nabs $21M wp.me/p8wLEc-bhRD by @Kyle_L_Wiggers
Joe Schmid retweeted
Katerina Wu, Penguins data scientist, on entering the sports analytics field: "This is perfect for me."
Read more about Wu, the first woman and Asian-American involved in scouting and evaluation for Pittsburgh's hockey operations staff: pens.pe/3de4Tku
Joe Schmid retweeted
If you’ve learned SQL and are trying to find a full time role in data or you are an analyst wondering what’s next... Learn @getdbt, find some data and open source your project. Far more companies hiring for dbt roles than there are candidates.
Joe Schmid retweeted
Basically my plan for how a bunch of open source science nerds can reshape humanity:
Step 1. Make @scikit_learn, @ProjectJupyter, etc.
Step 2. Get millions of people learning how to use ML & AI to augment their own wisdom, make better predictions
Step 3. Global prediction markets
Joe Schmid retweeted
"@CMU_Stats' Rebecca Nugent has been visionary and relentless — and I'm very excited to see her work recognized nationally." — Dean Richard Scheines on Nugent's recent honor from the National Academies (@theNASEM)
cmu.is/nugent #TartanProud
Joe Schmid retweeted
Also in 2006: @teoliphant released the first version of Numpy, unifying the array libraries in Python-land and starting a multi-decade revolution in scalable, vectorized computing for everyone.
It paved the way for modern machine learning and AI.
This quoted post is unavailable.
Joe Schmid retweeted
V0.16 of @blazingsql is out and we couldn't have done it w/out the continued support of an amazing ecosystem.
This release was heavily focused on bug fixes and quality of life improvements as we work on some major new features.
Thanks @rapidsai and everyone involved!
🔥🔥🔥 @RAPIDSai 0.16 is released (the blog will come in a week or so)! Great work to everyone in the ecosystem. And yes, @BlazingSQL is now a getting started selection, and will soon be a conda option. More big news soon about @RAPIDSai, @BlazingSQL, @dask_dev, and more! Woot!
Joe Schmid retweeted
“While we had initially built our own custom iterator for dealing with datasets larger than GPU memory, Dask-cuDF proved much more efficient, even on a single GPU” this is why @RAPIDSai used @dask_dev. Building bridges not walls.
With NVIDIA Merlin, you can build high-performing #recommenders at scale. Learn more about some of our latest updates to Merlin NVTabular, the Extract, Transform, Load (ETL) component of Merlin. #GTC20


























