Sladyn

Open source is your biggest internship 19-year-old me merged a tiny Jenkins Pull request and that commit snowballed into GSoC, the Linux Foundation, the Kubernetes release team, and eventually Meta.

@archiexzzz This is super interesting. I’ll try this out

At 18, I was obsessed with Neuroplasticity. I've been doing RL in my brain for 5 years now - training my amygdala and prefrontal cortex to become the person I wanna be. Through Neuroplasticity - you can rewire your subconscious thinking by repeatedly feeding your brain the right signal. Every time you repeat a thought or behavior, you're strengthening a specific neural pathway. Neuroscientists call this Hebbian potentiation - neurons that fire together, wire together. Do it enough, and it becomes your default way of thinking. I used Instagram for this: I saved a bunch of posts that reflect how I want to think. For me that's: > Virat Kohli > Larry Ellison > Steve Jobs > Travis Kalanick. I'm interested in how they "think". Your list can be completely different. Every single night, right before I sleep, I just watch through all my saved posts on repeat. The timing matters - your brain is in its most receptive state right before sleep which is when your hippocampus starts locking short-term impressions into long-term memory. I focus on three things: > what they say > how they say it > why they think like that. That activates your mirror neuron system - your brain starts rehearsing those patterns even just from watching. I kept doing this for 21 days straight. And I genuinely can't explain it well enough - I started thinking like them. Talking like them. Responding to situations like them. It changed something real in how I think. I think it'll do the same for you.

@_vmlops I need to feed this to Claude and make it train me

EFFICIENT LLM INFERENCE - The interview pocket notes you actually need drive.google.com/file/d/1mfTzOn…

Biggest game in 22 years on Sunday

Studying how data moves in the network during inference The movement of the KV cache across prefill and decode puts some strain on NICs The world of networking has come up with a great distinction between front end nics and backend nics

This is the future. Sharing context is a huge bottleneck in an organization Small teams 0 meetings only building Have the manager setup an LLM input where the reports make their LLM report their work in a day Move faster

the muse spark API will be coming soon! we have been thrilled with the amount of excitement amongst developers who want to try muse spark inside their agentic harnesses stay tuned!

cool to see people finding new emergent capabilities within Muse Spark!

North London X Meta

@TrapitBansal 🚀🚀

It’s been so much fun working with the extremely cracked team at Meta on this! This is just the start, even more capable models coming in the future. ai.meta.com/blog/introduci…

Liftoff 🚀

Building inference for Meta Muse Spark has been ridiculously fun. It honestly feels like we are just getting started Excited for the bigger challenges ahead.

1/ today we're releasing muse spark, the first model from MSL. nine months ago we rebuilt our ai stack from scratch. new infrastructure, new architecture, new data pipelines. muse spark is the result of that work, and now it powers meta ai. 🧵

What does this mean for penetration testers and red teams? I think we are entering a new generation of cybersecurity

I built a Claude skill for a team of engineers that I use to research execute and verify tasks. I let some of it run overnight and wake up to a bunch of actionable results

Studying RDMA. It’s actually such an interesting concept in order to reduce overhead we let the RNIC access the gpu memory directly

Alexa like device but with grok. I think you need to make it happen @elonmusk

Writing is the only time we learn and polish our cognition

Speculative decoding is such an interesting concept What do you mean you can just use a proxy model to generate tokens and use your main model for just verification