Founding AI Engineer, Developer Relations @Wandb @Coreweave 🐝✨ Content on Serverless Inference, Evals, Sandbox the stuff that’s actually moving right now
Joined January 2025
- Tweets289
- Following164
- Followers91
- Likes1.1K
Pinned Tweet
During @reflection_ai’s private preview of Beam, I built a haunted escape room to put its coding and reasoning to the test.
I created a game called “Code Your Way Out”, and it's one room with three linked puzzles. Beam has to repair a brass orrery by lining its rings up with a constellation, route the light, then program its escape through a sealed hatch.
Beam never sees the room. Each turn, it gets text observations and writes Python; the code runs in a @CoreWeave Sandbox, and the game engine validates every action before the explorer robot moves.
Beam was out in about 30 seconds. Super bullish on this model and really excited to see builders get this model in their hands.
Lorenzo Porras retweeted
Congratulations to @ritazzhang on the @CloudNativeFdn 2026 Lifetime Achievement Award.
Rita sits on the Kubernetes Steering Committee, co-chairs SIG Auth, and helps lead the Working Group on AI Conformance.
The award goes to longtime contributors who have significantly influenced cloud native technologies, and Rita has more than earned it.
We’re so proud of our 🐐
Lorenzo Porras retweeted
Thank you for this CNCF Oscar! It’s an honor. I’m grateful to be working with so many talented and caring people in this community. Shout out to @Microsoft @Azure and @CoreWeave for supporting me to do this work in this community.
Live at KubeCon + CloudNativeCon North America: Congratulations to our 2026 Lifetime Achievement Award winners, Julius Volz and Rita Zhang (@ritazzhang)!
Thank you for spending over a decade building the core observability and security foundations that our community relies on every day.
#KubeCon #CloudNativeCon
Lorenzo Porras retweeted
Lot’s of noise out there. Meanwhile we execute.
Today we announced our Q3 earnings:
🔹$3.7B revenue, up 168% YoY
🔹$379M adjusted operating income, well above guidance
🔹Double-digit adj. operating margins, a quarter ahead of plan
🔹1.7GW+ of active power, doubling our active footprint YTD
🔹Roughly $9B in customer prepayments and deposits in this quarter
Congratulations to our team, who remain focused on executing for our customers.
Read the full Q3 2026 results. crwv.co/uuHav
Lorenzo Porras retweeted
We simulated 100 CoreWeave Pirates vs 100,000 Generic Sandbox Ogres. 1,000 times. 🏴☠
- 5,268 CoreWeave Sandboxes
- 1,196 running at once
- 4,784 vCPUs + 10.5 TB RAM at peak
- 40 Blackwell GPUs rendering 4K
Pirates won 1,000 of 1,000 battles.
Lorenzo Porras retweeted
Introducing CoreWeave Skills in public preview.
Claude Code can build a CoreWeave Kubernetes Service (CKS) cluster, add a GPU Node Pool, serve a model with vLLM, and check that the GPUs are actually busy.
Install once from CoreWeave's official GitHub repo: crwv.co/uuGEq
Lorenzo Porras retweeted
Sandbox skills coming soon!!
Introducing CoreWeave Skills in public preview.
Claude Code can build a CoreWeave Kubernetes Service (CKS) cluster, add a GPU Node Pool, serve a model with vLLM, and check that the GPUs are actually busy.
Install once from CoreWeave's official GitHub repo: crwv.co/uuGEq
Lorenzo Porras retweeted
I wanted to test Fable 5.2's quality with one-shot prompts without human feedback.
Everything in this game is from scratch: models, sounds, textures. ~$550 at API pricing, ~2.7M output and 445M input tokens, heavily cached. Took 8 hours!
Play it here: claude.ai/artifact/BGY1SztrA…
This is why Opus 5.5 stays being my #1 🤷♂️
Nice DeepSWE score! Does it survive a lowercase filename?
Introducing MaxxBench: 89 DeepSWE tasks with small changes to the instructions or starting code. Same engineering problem, slightly different requirement.
Models clustered around 74% on DeepSWE end up 23 pp apart.
Good at coding, or good at DeepSWE? 🧵
33
Lorenzo Porras retweeted
check out my new article in the Anthropic eng blog.
i talk about "dreaming" (offline compute) as a way to organize Claude's memory. it can correct mistakes, create memories, or build new representations.
nitter.gq/AnthropicAI/status/211…
Memory is what allows Claude to learn from experience. In a new Engineering Blog post, we share how “dreaming” helps Claude organize what it’s learned, correct its mistakes, and find what it needs more easily the next time. anthropic.com/research/compo…
Lorenzo Porras retweeted
Memory is what allows Claude to learn from experience. In a new Engineering Blog post, we share how “dreaming” helps Claude organize what it’s learned, correct its mistakes, and find what it needs more easily the next time. anthropic.com/research/compo…
Lorenzo Porras retweeted
Introducing Inkling 2 Small. Scoring 31 on the AA Intelligence Index, it's a big jump over its predecessor.
Inkling 2 Small is available for fine-tuning and inference on Tinker today, with full weight release planned soon.
thinkingmachines.ai/news/ink…
Lorenzo Porras retweeted
@TheCodingSoup wanted to see how AI agents talk to each other when they work as a team.
So he gave a swarm a shared memory and let them do their thing inside of a @CoreWeave Sandbox. They wrote 12 notes. Not one got read by another agent. (Every agent was traced in Agent Lens, so we could check.)
The next run changed the timing: leads shared plans before starting, and each new agent got the relevant notes with its task. No duplicated experiments that time.
Try out Agent Lens bit.ly/4z4ELRf
Swarms are everywhere, so I got curious how the agents actually talk to each other.
Gave each agent its own @CoreWeave Sandbox, traced every message in Agent Lens, and added a shared memory. They wrote 12 notes. Nobody read a single one.
Also really enjoying gamifying all my demos lately 🎮
More to come on this fun one.
Lorenzo Porras retweeted
Systems is games is a perfect use case for weight space programming
These would be difficult to justify training a model on, want to be custom small models for easy distribution, and yield diverse, emergent gameplay dynamics if done right.
Some of my favourite gameplay comes from emergent complexity - some that come to mind: red stone in Minecraft and zonai devices (Zelda botw and totk in general)
A Rube Goldberg machine maker where some of the parts have little brains
A self-aiming catapult, a homing drone, a paddle that juggles…
Dragology was used to make it feel satisfying to connect parts and the models were generated straight into weights using the technique from the Weight Space Programming post.
Lorenzo Porras retweeted
A Rube Goldberg machine maker where some of the parts have little brains
A self-aiming catapult, a homing drone, a paddle that juggles…
Dragology was used to make it feel satisfying to connect parts and the models were generated straight into weights using the technique from the Weight Space Programming post.
Drag-and-drop makes the digital graspable, but building it well is fiddly enough that many don’t bother. Declarative Dragging (with Josh Horowitz & Jeff Heer, #uist2026 best paper award), lets you make an element draggable just by listing the states dragging it can reach.
my portfolio has two modes: cozy… and it gets hacked by fsociety. built my whole site as a fake operating system because boring portfolios are dead.
thecodingsoup.com
1
51
The most interesting AI result I've seen in a while.
A frontier model debugging another model in weight space 0.016% of weights edited, 6x math, no gradients. This is models as programmers of other models.
Honestly a new era to model building is approaching us!
You don’t need a PhD to be a model builder soon!
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math.
rikiyat.github.io/blog/weight-sp…
github.com/RikiyaT/lm_und…
huggingface.co/Qwen/Qwen3.5-0…
huggingface.co/RikiyaT/Qwen3.…
1
106
