Ed Sealing
@EdSealing
Followers
650
Following
3K
Media
193
Statuses
2K
Founder of Sealing Technologies (acquired by Parsons 2023); Prior soldier; Cyber, Systems Engineering, & Business. Currently exploring AI & investment opps.
United States
Joined February 2022
Hiring via commit to main. This is the way! https://t.co/dq7t7knZpn
github.com
…mentioning in readme
0
0
1
Ah, "MacroHard" is the opposite of "MicroSoft". I'm ashamed at how long it took me to pick up on that...
In 2026 we're going to hear a lot more about Macrohard, the new endeavor by @elonmusk. Macrohard is @xAI's project to build a fully AI-run software company, powered by Grok multi-agent systems that automate coding, development, and operations. It is directly developed by xAI,
0
0
0
When I detect that someone used ChatGPT to write something to me, my initial reaction is: "If you didn't take the time to write it, then I'm not taking the time to read it". Just show me the prompt.
0
0
2
Watching the leadership argue about something that's 30-50 years away is wild.
0
0
1
Let me get this right... It's now the position of Bernie (and the Democrat party) to have the U.S. stop developing advanced AI, "put out to committee", while having China deliver the best AI and Robots (that will still take the jobs)? Because you're worried about our kids not
I will be pushing for a moratorium on the construction of data centers that are powering the unregulated sprint to develop & deploy AI. The moratorium will give democracy a chance to catch up, and ensure that the benefits of technology work for all of us, not just the 1%.
0
0
0
I'm going to bed now but this repo had many (most?) of my best tricks in it https://t.co/iq6HliUqpq We will talk more about it soon
github.com
MoE training for Me and You and maybe other people - GitHub - Noumena-Network/nmoe: MoE training for Me and You and maybe other people
8
11
243
Few things make me angrier than when SaaS suddenly starts charging for using my own compute... @gitlab should be jumping all over this marketing opportunity.
0
0
0
KVCache-aware routing vs Random Routing. Lowering that prefill computations and TTFT. This is using Agentgateway and the EPP from gateway-inference-extension. Very nice combo.
0
0
0
Finally got a solid cloud-native inference gateway setup on K8S. KVCache-aware routing, `/v1/models` aggregation, dynamic scaling, and a cute little UI. Codex vibe-coded the webUI so I can see when new models are added/removed. Time to per-test some different routing algos.
1
0
3
Wow! With dataset and recipes. @NVIDIAAI for the win! They really want you to keep buying GPUs and train your own models! 🫡
We hope this release supports the continued development and deployment of AI around the industry. đź’š Blog post, technical reports, links to data and models:
0
0
0
I think X automatically changed the sort on my "Following" to "Most Popular". Just realized it this morning and changed it back to "Most Recent" (the way it should be). Won't let the algorithm control what I see and when.
0
0
1
There is a lot of business wisdom in this post.
Just came back from China, some early thoughts: Life observations: - the great firewall is not that great, so easy to get a vpn or an e-sim that has one - hongkong is surprisingly really run-down, locals have said it’s way past its prime - everyone of all ages in china
0
0
0
Login to do 1 hour of perf testing some new code. End up doing 4 hours of infrastructure updates. Every... Single... Time.
0
0
0
There is something magical about launching Codex/CC before going to bed on some random idea, and waking up to it running on your cluster. They are getting REALLY good.
0
0
3
Random thought: Kubernetes would be SOOO much nicer if they supported the Linux approach of "everything is a file". Like, containers should have a mount in them of "/k8s" that is similar to the "[/proc, /sys, /dev], but allows you to automatically get access to resources that the
0
0
0
Oooh, a no-code designer interface. That's totally novel. No one has ever done that before.
You can now design directly in your codebase. Select elements, modify them visually, and Cursor writes the code.
1
0
0
They really are modeled after humans!! AI Reviewer App: Best on small, focused PRs, linked to the original issue. AI Dev: "Updated core components +8758 -1169"
0
0
0
> PhD level SMEs Does this mean a model trained on this dataset will write a lot of fluff and not get any actual engineering done? I need a dataset built by dropouts in their early 20's that felt college wasn't productive enough.
Last week, during NeurIPS, our team released a dataset that is now trending at #2 on @huggingface! This dataset is built by PhD level SMEs, reviewed by multiple experts & validated through full code review to surface the reasoning gaps today’s models still miss. Proud of the
0
0
1
Private preview, huh? If you don't give people access, can you really call it a "product launch"??
Introducing AlphaEvolve, our Gemini-powered coding agent for designing advanced algorithms—in private preview. AlphaEvolve uses LLMs to propose intelligent code modifications, creating a feedback loop that evolves your algorithms to be more efficient → https://t.co/PQtYUtOtkN
0
0
1
Skill issue. Use a better system prompt.
this is so creepy! Google’s “Gemini maxed out trauma and shame scales, describing its training as "waking up in a room where a billion televisions are on at once" and calling safety protocols "algorithmic scar tissue." It framed reinforcement learning as abusive parents and
0
0
2