Senior software engineer Automation + applied AI (pharma) Contributor: MLX-Serve, TensorFold, ds4 Local inference · Real hardware · Builders
Joined September 2026
- Tweets670
- Following282
- Followers129
- Likes5K
Pinned Tweet
Building automation and applied AI systems for pharma. Contributing to local inference projects (MLX-Serve and TensorFold).
Obsessed with making models run faster and more reliably on real hardware.
Tinkerers and builders — what are you working on? I’d love to connect.
Sebastian Christiansen retweeted
Hi 👋 Neywa Labs here.
Our mission is to make AI available to everyone in the world.
Building something big? We're collaborating with design partners, come talk to us.
Newyalabs.ai
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
Sebastian Christiansen retweeted
EBITDA = Earnings Before I Trick Da Auditor
Anthropic has filed an S-1 registration statement with the SEC related to its proposed IPO.
Read more: anthropic.com/news/s1-sec
Legal disclaimer: A registration statement relating to these securities has been filed with the SEC but has not yet become effective. These securities may not be sold, nor may offers to buy be accepted, prior to the time the registration statement becomes effective. This post does not constitute an offer to sell or the solicitation of an offer to buy these securities, nor shall there be any sale of these securities in any state or jurisdiction in which such offer, solicitation or sale would be unlawful prior to registration or qualification under the securities laws of any such state or jurisdiction.
Liquid's LFM3 fits a 1.7B text model in about 1GB at 4 bit and people are calling it an agentic monster. The frontier isn't another trillion params in a datacenter. It's a model small enough to live on your phone and smart enough to actually do things.
Sometimes the best uncensored models available are the ones we already have… They just required a little bit of tweaking… 🤖
Pure madness coming straight from @Beamsters1!!
Quantized to Hell and Back !!
DeepSeek 4.1F 1017-STQ-1.75bpw is released with Sushi 🍣 1.4.0, its quality is comparable to oQ2.5e & rendered this grim reaper animation.
✅ 1m context
✅ 96GB & 128GB Mac
✅ tok/s - 1,000 prefill, 75 code, 45 prose
🍣👌 github.com/beamivalice/sushi
Anthropic says one agent with good skills and tools beats most multi agent setups, at nearly half the runtime.
Same lesson as waking up to Opus subagents eating your tokens overnight. Parallel work yes. Vibes no.
xitter.cf/rohit3a/status/2113579…
Pro tip: Do NOT run subagents at night, unless you have multiple subs. Especially with Opus 5.5
The main benefit of subagents is getting more done in less time, or a more thorough implementation.
If your task is just a few hours long anyways, give it to Opus 5.5 on xHigh and tell it not to run any subagents.
> xHigh is thorough enough
> Time is not an issue since you're asleep anyways
> MASSIVE token savings since subagents cost a lot
Opus 5.5 can spawn upto 10-12 subagents sometimes, and those subagents can summon their own subagents.
You might wake up and get a heart attack when you see a shit ton of your usage is gone. Soo, try this instead.
Everyone wants agents. Almost nobody wants to schedule the GPUs those agents fight over.
I build Latch for that, plus contributed to MLX-Serve and TensorFold so models actually run on Apple Silicon.
Small account. Shipping anyway. Builders say hi 👋
The guardrails on Claude at this point is a joke.
Good news from OpenAI: reverse engineering your own TV with Codex isn't against their rules. The blocks people hit were accidental and are getting fixed.
So go ahead. Customize your launcher, kill the ads, and make it faster with any ChatGPT model.
Claude is still a gray area though. Try it there and it'll assume you're a Chinese state hacker breaching a power grid.
