Vivek Raghunathan (@vivek7ue) 's Twitter Profile
Vivek Raghunathan

@vivek7ue

* SVP eng at @snowflakedb.
* AI + search at @snowflakedb.
* Co-founder @Neeva . #NeevaAI = AI search.
* Ex-VP of Eng @Google (ads,YT, Google Now)

ID: 815659822435606528

linkhttps://www.linkedin.com/in/raghunathanvivek/ calendar_today01-01-2017 20:43:50

984 Tweet

4,4K Followers

1,1K Following

Casper Hansen (@casper_hansen_) 's Twitter Profile Photo

Almost a 5x speedup in vLLM🤯 I was able to push a finetuned Mistral Nemo from 110 tokens/s to a peak of 517 tokens/s and acceptance rate of 57.7%. This is with Suffix Decoding from ArcticInference⚡

Dwarak Rajagopal (@dwarak) 's Twitter Profile Photo

Exciting news! The PyTorch Foundation’s expansion with vLLM and DeepSpeed is a game-changer for open-source AI. Can’t wait to see the innovations this brings! As a premier member, Snowflake is excited to join the Board and help grow the OSS community. Big things ahead! 🚀

Snowflake (@snowflakedb) 's Twitter Profile Photo

Anthropic’s cutting-edge Claude Opus 4, and Claude Sonnet 4 models are now available to customers on Snowflake Cortex AI. We're thrilled to build on our strong partnership with Anthropic, bringing their latest innovations for easy, efficient, and trusted AI to enterprises.

Anthropic’s cutting-edge Claude Opus 4, and Claude Sonnet 4 models are now available to customers on Snowflake Cortex AI.

We're thrilled to build on our strong partnership with <a href="/AnthropicAI/">Anthropic</a>, bringing their latest innovations for easy, efficient, and trusted AI to enterprises.
Yusuf Ozuysal (@yusufozuysal) 's Twitter Profile Photo

How do faster inference (up to 16x for embedding models!) and better Text2SQL through RL sound? A jam-packed launch from Snowflake AI research team detailing the technologies bundled in our ArcticInference framework and also diving deeper into how the model at the top of the

Jeff Rasley (@jeffra45) 's Twitter Profile Photo

🧵1/ New release from Snowflake AI Research: Shift Parallelism is a new LLM inference technique built on top of vLLM, released through ArcticInference. It dramatically improves latency while preserving high throughput. Here’s what it looks like in action 👇

Stas Bekman (@stasbekman) 's Twitter Profile Photo

In inference one usually gets either high throughput or low latency, but not both - enter shift parallelism which automatically adapts for the best performance!

sridhar (@ramaswmysridhar) 's Twitter Profile Photo

Just announced: Snowflake has agreed to acquire Crunchy Data, bringing open-source Postgres tech into the AI Data Cloud. With this news, we will be introducing Snowflake Postgres: enterprise-grade, AI-ready, and fully managed. Run your most critical, AI-powered apps on

Just announced: <a href="/Snowflake/">Snowflake</a> has agreed to acquire <a href="/crunchydata/">Crunchy Data</a>, bringing open-source Postgres tech into the AI Data Cloud.

With this news, we will be introducing Snowflake Postgres: enterprise-grade, AI-ready, and fully managed.

Run your most critical, AI-powered apps on
Crunchy Data (@crunchydata) 's Twitter Profile Photo

We are excited to announce that Crunchy Data is joining Snowflake to bring Postgres to the AI Data Cloud. 🎉 crunchydata.com/blog/crunchy-d…

Vivek Raghunathan (@vivek7ue) 's Twitter Profile Photo

🚨 Big news at #SnowflakeSummit! 🎉 Snowflake acquiring @CrunchyData to bring the power of OSS Postgres to the AI Data Cloud 🚀 Say hello to Snowflake Postgres — enterprise-grade, AI-ready, and fully managed Welcome, team Crunchy! Let’s build the future of data together!🔥

Prashanth Chandrasekar (@pchandrasekar) 's Twitter Profile Photo

Today we're excited to announce that Stack Overflow Knowledge Solutions will now be available on the Snowflake Marketplace! Armed with the wealth of trusted, accurate knowledge from Stack Overflow's over 16 years worth of data, customers can now easily enrich their AI

Today we're excited to announce that <a href="/StackOverflow/">Stack Overflow</a> Knowledge Solutions will now be available on the <a href="/Snowflake/">Snowflake</a> Marketplace! Armed with the wealth of trusted, accurate knowledge from Stack Overflow's over 16 years worth of data, customers can now easily enrich their AI
Snowflake (@snowflakedb) 's Twitter Profile Photo

Data teams are going through a big shift right now towards Agentic Analytics Systems. In just the last year, we’re seeing autonomous LLMs, unstructured business data systems, and a shift to business users being able to do their own analysis. #SnowflakeSummit

Data teams are going through a big shift right now towards Agentic Analytics Systems.

In just the last year, we’re seeing autonomous LLMs, unstructured business data systems, and a shift to business users being able to do their own analysis. #SnowflakeSummit
Snowflake (@snowflakedb) 's Twitter Profile Photo

Our research team recently launched a vLLM plugin, Arctic Inference, utilizing a slew of in-house built technologies like SwiftKV, speculative decoding, suffix decoding for blazing fast speeds. With these inference optimizations, enterprises can see up to 4x improvements on

Our research team recently launched a vLLM plugin, Arctic Inference, utilizing a slew of in-house built technologies like SwiftKV, speculative decoding, suffix decoding for blazing fast speeds. With these inference optimizations, enterprises can see up to 4x improvements on
Vivek Raghunathan (@vivek7ue) 's Twitter Profile Photo

There is a lot of buzz about AGI but not many specifics.  Join Jared Kaplan from Anthropic in 38 mins at Snowflake summit to hear the roadmap to human level capabilities. At Moscone today. Register free at: snowflake.com/en/summit/dev-…

Stas Bekman (@stasbekman) 's Twitter Profile Photo

My first project at Snowflake AI Research is complete! I present to you Arctic Long Sequence Training (ALST) Paper: arxiv.org/abs/2506.13996 Blog: snowflake.com/en/engineering… ALST is a set of modular, open-source techniques that enable training on sequences up to 15 million

My first project at <a href="/Snowflake/">Snowflake</a> AI Research is complete! 

I present to you Arctic Long Sequence Training (ALST) 

Paper: arxiv.org/abs/2506.13996
Blog: snowflake.com/en/engineering…

ALST is a set of modular, open-source techniques that enable training on sequences up to 15 million
sridhar (@ramaswmysridhar) 's Twitter Profile Photo

Snowflake has achieved FedRAMP High authorization on Microsoft Azure Government. This is a significant milestone, reinforcing our commitment to empowering U.S. government agencies with the most secure and scalable data and AI capabilities. We're proud to help accelerate

Snowflake (@snowflakedb) 's Twitter Profile Photo

We’re excited to be a launch partner with OpenAI as they release their open‑source models, available today natively in Snowflake Cortex AI. OpenAI’s state‑of‑the‑art gpt-oss open‑weight models deliver strong real‑world performance at low cost. Available on Snowflake Cortex AI

We’re excited to be a launch partner with <a href="/OpenAI/">OpenAI</a> as they release their open‑source models, available today natively in Snowflake Cortex AI.

OpenAI’s state‑of‑the‑art gpt-oss open‑weight models deliver strong real‑world performance at low cost. Available on Snowflake Cortex AI
Snowflake (@snowflakedb) 's Twitter Profile Photo

AI Observability in Snowflake Cortex AI is now generally available! 👩‍💻 Developers can effortlessly evaluate & trace Gen AI apps, boosting reliability & trust. It helps assess performance, debug, & refine AI systems. Built-in across Cortex features for end-to-end control. ❄️

AI Observability in Snowflake Cortex AI is now generally available! 👩‍💻 

Developers can effortlessly evaluate &amp; trace Gen AI apps, boosting reliability &amp; trust. It helps assess performance, debug, &amp; refine AI systems. Built-in across Cortex features for end-to-end control. ❄️
sridhar (@ramaswmysridhar) 's Twitter Profile Photo

. OpenAI's GPT‑5 is now live in Snowflake Cortex AI. Build AI apps. Analyze multimodal data. Launch intelligent agents. All natively where your data lives with no integration hassle and consistent data and AI governance. The future of enterprise AI is here. Let’s build. 👇

Stas Bekman (@stasbekman) 's Twitter Profile Photo

Yay, our team has just published a new paper, “Shift Parallelism: Low-Latency, High-Throughput LLM Inference for Dynamic Workloads" arxiv.org/abs/2509.16495 Shift Parallelism is a new inference parallelism strategy that can dynamically switch between Tensor Parallelism and

Yay, our team has just published a new paper, “Shift Parallelism: Low-Latency, High-Throughput LLM Inference for Dynamic Workloads"

arxiv.org/abs/2509.16495

Shift Parallelism is a new inference parallelism strategy that can dynamically switch between Tensor Parallelism and