Ronen Dar

Co-founder & CTO at Run:ai

[theCUBE @theCUBE]

Tune in now to #theCUBE’s real-time broadcast from #SC24 as we speak with @ronen_dar about how @runailabs collaboration with @weka support AI demands of enterprises.

📺 Catch the complete discussion live!
https://t.co/CenZRKdlCY

#Supercomputing #HPC #LiveAnalysis
@savissavvy

[Ronen Dar @ronen_dar]

Blog:
https://t.co/if7HWd3dhe

Benchmarking paper: https://t.co/On1Xumh9Nq

[Ronen Dar @ronen_dar]

We've released Run:ai Model Streamer as open source to help the community load inference models faster than ever before 🎉

This is part of our efforts at @runailabs to make inference workloads more dynamic and cost efficient.

GitHub repo:
https://t.co/BrOK2Chi6m

[Ronen Dar @ronen_dar]

How Run:ai Streamer helps?
- 6x faster model loading times
- Works seamlessly across all storage types (local, S3, NFS)
- Native integration with popular inference engines like @vllm_project
- Zero model format conversion needed to allow seamless integration with @huggingface

[Ronen Dar @ronen_dar]

Making Data Simple
https://t.co/UING6yMFzA

[Ronen Dar @ronen_dar]

I had the pleasure of being a guest on a couple of great podcasts recently 🙌

Shared thoughts about what we’re doing at @runailabs and about the advancements happening in the AI and GPUs world

Check out the episodes linked below

[Ronen Dar @ronen_dar]

The Data Driven Podcast
https://t.co/FQnVUol02B

[Ronen Dar @ronen_dar]

Omri and I speak about 2023 and plans for 2024 😀💪

https://t.co/wN0virUSpX

[DataCamp @DataCamp]

Discover the challenges and best practices of deploying LLMs into production environments.

@ronen_dar, Co-founder @runailabs, explores optimizing AI model training to reduce costs, facilitating the adoption across diverse business scales, and more.

🔗👇 https://t.co/Jx4TaSF6lC

[Heather Joslyn🇺🇦 @ha_joslyn]

Standing room only for @ronen_dar
of Run:ai at #kubecon #cloudnativecon co-located talk on training #LLMs on #Kubernetes. #AI needs dynamic compute but it’s hard to have GPU access when needed when GPU is shared manually.

[Ronen Dar @ronen_dar]

@sharongoldman writes about how @runailabs and other Israeli AI startups are carrying on after the horrors of the last few days.

This is the biggest disaster Israel has ever had but as mentioned in the article “we’ll prevail and become stronger” 🇮🇱

https://t.co/SQLzF6K6JF

[Ronen Dar @ronen_dar]

Thanks @mlopscommunity and @RahulParundekar @adamboazbecker for organizing this great event!

I had awesome time speaking and listening to great talks from @llama_index and @NimbleBoxAI

AI is changing the world and the energies behind it feel absolutely amazing

[Ronen Dar @ronen_dar]

Going to speak today at the San Francisco MLOps Community Meetup 😀

Hope it will be fun and interesting !

https://t.co/m6lAbMZ0Zl

[Ronen Dar @ronen_dar]

"Given that GPU availability is poised to remain a challenge for the foreseeable future, product leaders must think strategically about GPU allocation."

✅

[VentureBeat @VentureBeat]

A new way to optimize and prioritize AI projects for the GPU shortage https://t.co/esGudISmHd

[Ronen Dar @ronen_dar]

Teams in Meta fight internally on access to GPUs

We’ve heard this story countless times from Run:ai’s prospects 🤦‍♂️

Apparently, GPU allocation is a huge problem even in “GPU-rich” companies

https://t.co/3ZZmMArRNb

[Ronen Dar @ronen_dar]

Scheduling distributed computing workloads on shared CPU Clusters is not trivial

More about Run:ai’s Kubernetes CPU scheduling over here:
https://t.co/DVUFT3vWjT

[Amir Efrati @amir]

Exclusive:

OpenAI is blowing by earlier revenue projections, already passed $1b ARR.

Turns out plenty of big companies like LLMs :)

https://t.co/4ei2UMIqJx w / @aaronpholmes @OpenAI $msft

[Ronen Dar @ronen_dar]

@julien_c I think you're wrong :)

1000 GPUs today is low...