The MLOps community is an open and transparent community where all are welcome to participate. It is a place where MLOps practitioners can collaborate and share
Joined March 2020
- Tweets 5,711
- Following 398
- Followers 11,318
- Likes 3,538
1,240 Photos and videos
Pinned Tweet
Coding Agents Conference. March 3rd at the Computer History Museum. luma.com/codingagents
3
12
3,158
2⃣ Most of today's agent products solve for one person working with one agent. Stan thinks the harder problem is building a workspace where multiple humans and multiple agents can hand work off without everything falling apart.
2
78
3⃣ Stan also argues that even if model progress slowed down tomorrow, AI would still fundamentally change how we work. The bigger opportunity may be in the systems around the models rather than the next model release.
go.mlops.community/tc3yyt
83
As the MLOps Community becomes @AgenticAIFdn's user group, we're launching Agentic Conversations- a new podcast and video series with the builders, researchers, and practitioners working on agentic AI systems.
2
1
3
303
Less agent hype. More conversations with the people building, testing, and learning in practice.
Watch the trailer 👇open.spotify.com/episode/2ck…
1
123
Catch the first 3 episodes 👇!
📺 YouTube: youtube.com/@demetriosbrinkm…
🎧 Spotify:
open.spotify.com/show/033rZZ…
124
What if the biggest risk in your AI stack isn't the model...
...it's retrieval?
As agents gain access to more data and more tools, the line between helpful automation and unintended access gets blurry fast.
1
1
1
262
Join this week's MLOps Reading Group as we discuss Securing the Agent: Vendor-Neutral, Multitenant Enterprise Retrieval and Tool Use.
We'll explore why enterprise AI security may need to be rethought from the ground up.
📅 Tomorrow, June 11
🕜 1:30 PM ET
1
1
141
Bring your questions, skepticism, and hot takes.
Register here now: home.mlops.community/home/ev…
127
Just listened to James Everingham (@GuildAI, previously Meta/Instagram/Netscape) talk with Demetrios about what it takes to run agents inside a company without them wrecking things.
1
194
3⃣ Source control wasn't built for this. When a human and an LLM co-author the same PR, line-level diffs don't cut it - he wants character-level provenance, partly so a regulated shop can answer "what did this agent do at 2 pm three years ago."
1
61
go.mlops.community/he3hev
Curious where people land on the one-big-agent question — anyone actually running the orchestrated many-small-agents setup in prod, or is it still cleaner to let one capable agent own the whole flow?
65
🚨 Real-time ML is hard. Feature freshness, low-latency inference, training-serving consistency, streaming pipelines—there's a lot that can go wrong.
1
2
372
Join engineers and ML practitioners from @Intuit, @creditkarma, @depop, and @OpenAI as they share how they're using Zipline and Chronon to build real-time ML systems in production.
1
387
📅 June 10
🕤 9:30–10:30 AM PT
💻 Virtual
🎟️ Register today:
home.mlops.community/public/…
128