Got plans for Nov 3? We'll be at AWS Loft in SF for a hack night.
We'll build a LangGraph agent that tries open models, fine-tunes promising candidates, and deploys the winner, all in your own AWS account.
Come spend an evening building with us!
Nov 3rd at AWS Loft in SF: Own Your AI: Build an Agentic AI Model Factory - Hack Night
We'll start with a walkthrough where we build a LangGraph-powered agent that evaluates open models in parallel, fine-tunes promising candidates, deploys the winner, and retrains automatically when performance drops.
Along the way, you'll see how Union.aii gives your workflows a durable runtime with infrastructure as context, so agents adapt infra resources, recover from failures, and keep moving without re-running completed work.
And because it all runs in your own AWS account, you own the full stack end to end: your models, your data, your infrastructure, your IP.
Think of it as the first brick in your own model factory: a repeatable system for evaluating, improving, and deploying models you control!
Excited to see people there!
Your queue allows 100 jobs. Each needs 8 GPUs. Your cluster has 16. Enjoy the pending pods.
Part 4 of Inside Union covers how we added resource-aware scheduling to Flyte 2, checking demand against quotas and cluster capacity before dispatch.
Read the post: union.ai/blog-post/inside-un…
That AI project you keep saying you'll get around to? Bring it tonight 👀
We're hacking on sovereign AI factories at our Bellevue office with @sagecodes, @niels_ai and @GalarnykMichael. Try open models, build agents, experiment with evals or bring your own project.
RSVP here: luma.com/9ngtvw3u
We're hosting another hacknight this Thursday (Oct 22nd) at the @union_ai Bellevue office: This months topic is owning your AI by build agentic model factories!
@niels_ai and @GalarnykMichael are in town and will be there as well!
You'll see how Union ai & Flyte give your AI workflows a durable runtime with infrastructure as context, so agents adapt infra resources, recover from failures, and keep moving without re-running completed work. And because it all runs in your own cloud, you own the full stack end to end: your models, your data, your infrastructure, your IP.
Then we switch into hack night mode: keep building on the example, bring your own models or datasets, experiment with evals and fine-tuning, extend open models to different parts of your agent, or start something new around the same ideas.
Should be fun and i hope to you there!
Using a hosted model shouldn't mean giving up control. Running your own shouldn't mean giving up your development speed.
That's our take on sovereign AI: you choose where each workload runs 🧭 and Union keeps it durable as you scale 💪
We're in! Union is now part of the @CoreWeave Partner Network 🤝
Building an AI factory? Bring your Python. Union connects data prep, training, evals and inference on a durable AI runtime, on infrastructure you control 🔒
coreweave.com/coreweave-part…
Follow the curve to Booth 45 👋
We’ll be at #PyTorchConf in San Jose, Oct 20-21, showing distributed training on Union and how to keep an eye on GPU health and utilization in your own cloud.
Come say hi to @niels_ai and @cre8minus1!
Union AI retweeted
💥 A 9-hour AI pipeline fails at hour 8. Whether you pay for the first 7 hours twice comes down to which verb you reach for next.
Flyte 2 has three, and they differ on whose code runs and what re-executes.
"I waited too long to switch to Flyte 2."
That's @PrimaMente's @GanbatJay, after moving a production pipeline in about half a day. Their largest run on Union hit 278,000 actions, in plain Python.
If you're still working around your orchestrator's limits, this is worth a read 👇
How Prima Mente scaled DNA foundation model pipelines past 250,000 tasks: union.ai/case-study/prima-me…
Good writeup on Anthropic's dynamic workflows: Claude now plans the work, writes the orchestration, and spins up its own agents.
The detail we'd underline: these runs cost a lot more than a single agent. Once agent runs look like that, a crash 80% of the way through can't mean starting over.
latent.space/p/claude-dynami…
This is the problem we work on at Union.ai: durable runtimes where failed steps recover without re-running completed work.
At AI Engineer NYC this week? Come find us at booth #48.
NYC, who's building cool things with AI? 👀
The Union team will be at AI Engineer this week!
📅 Oct 12–14
📍 Booth 48
Come meet the team and see what building sovereign AI on your own cloud looks like!
@sagecodes @cre8minus1, see you there? 👋
Union AI retweeted
Today we're launching the Alzheimer's Translation Challenge, a global AI competition to discover new therapeutic hypotheses for Alzheimer's disease.
At its core is a new 150M cell atlas of neurons, astrocytes, and microglia across different genetic backgrounds under combinatorial perturbations, with multi-modal readouts.
The data will be made available through the AD workbench, Hugging Face and Prima Mente's modeling platform.
The challenge is led by @PrimaMente and @AlzData in collaboration with @nvidia, @huggingface, @nebiusai, Talisman Therapeutics, @UltimaGenomics, @cellanome, @PrimeIntellect, @GoodfireAI, and @boltz_bio.
In 1906, Alois Alzheimer described a new disease of the brain. After 120 years of progress, we still don't fully understand what drives this disease. We want to change that.
Participants will use the dataset to build AI models, assessed through a series of evals from general benchmarks to more complex tasks.
Top teams will be shortlisted, and their experimental hypotheses tested in Prima Mente's wet lab. Teams win when their hypotheses demonstrate the predicted molecular and functional effect.
A question we get: "Does our data go through Union?"
Your inputs, outputs, logs and code stay in your cloud account. Even when you open logs in our console, your browser fetches them from your infrastructure.
Union AI retweeted
This weeks Flyte log: Run Claude Agent SDK Agents on Flyte, With Every Tool as a Durable Task
Agent frameworks are good at deciding what to do next. They are less good at what happens when the process dies, when one tool needs a GPU and another needs 200 MB of RAM, or when someone asks what the agent actually did in a past run.
flyteplugins-agents-claude splits that work. You keep writing agents with the Claude Agent SDK. Flyte becomes the durable runtime underneath: every agent, sub-agent and tool call becomes a containerized durable Flyte task with its own resources, retries, and cache, and the whole conversation renders as a timeline in the task report.