Twitxy Saved

Aditya Parameswaran

Verified account This account is verified. Learn more

@adityagp

associate prof @ucberkeley, ai x data research lead @snowflake, co-director @ucbepic, cofounder @ponderdata (acq. @snowflake)
Berkeley, CA adityagp.net Joined April 2008
779 Following 5.7K Followers 2K Posts

Latest posts 5

Aditya Parameswaran Verified account @adityagp
A big *thank you* to @NSF/Stride Ventures for the support for our efforts on improving cost and quality for large-scale AI-powered data processing workloads - and deploying it with industry partners. Excited to work with @sh_reya and @profjoeyg on this, and that we're one of two teams with @Berkeley_EECS origins, out of 9 that received the award!

We're always hiring PhD students and postdocs - apply to Berkeley and also to @CMUDB (to work with @sh_reya)!

Here's more: stride-ventures.com/start2-groups-…
Shreya Shankar Verified account @sh_reya Oct 22
Very excited to share that we've been selected to receive NSF funding through the NSF-STRIDE AI Efficiency Challenge!

People are increasingly using LLMs to analyze large datasets, but doing so with current AI models and systems is often extremely expensive, slow, and inefficient. We need better infrastructure to make this kind of analysis practical at scale.

I am hiring PhD students and postdocs to work in this area starting in 2027—prospective PhD students, please apply at the CMU website, and prospective postdocs, please send me an email.
View quoted post on X
0 replies 3 reposts 21 likes 1.9K views View on X
Aditya Parameswaran Verified account @adityagp
The actor-gatherer design is what we believe the future of agentic knowledge work should look like - decoupling reasoning (act) from proactive context management (gather) - to handle long-running, parallel conversations with agents along with capturing changes to the environment. The framework is bolt-on to any coding agent. This is the logical conclusion to many context management proposals - compaction, RLMs, CLMs most of which conflate gathering and acting - and the accuracy improvements are impressive as a result. Try it out in your favorite harness and let us know what you think!
Sepanta Zeighami @SepantaZeighami Oct 20
1/6 🚨Releasing Actor-Gatherer framework, a new paradigm for agent context management developed at UC Berkeley for working with multiple agents in parallel over long sessions. Actor-Gatherer decouples knowing from acting: Actor agents perform tasks and Gatherer sets their context
View quoted post on X
4 replies 3 reposts 41 likes 2.6K views View on X
Aditya Parameswaran Verified account @adityagp
All existing AI-SQL systems focus on optimizations at the logical + physical level, leaving the inference stack alone - some exciting new work from @sh_reya and @charles_irl shows there is lots of low hanging fruit in that layer. Awesome!
Shreya Shankar Verified account @sh_reya Sep 24
Excited to share Quail, our new open source AI-SQL engine (a collab with Modal)! By planning queries and LLM inference together, it reaches 1B+ input tokens/min on one H100 for one query 😱🚀

AI-powered data operators create a new, interesting inference workload👇
View quoted post on X
2 replies 1 reposts 10 likes 2.7K views View on X
Aditya Parameswaran Verified account @adityagp
Couldn't be more excited to join @Snowflake part-time to lead AI x Data research — and help envision how AI reshapes data infrastructure, broadly defined!
Anupam Datta Verified account @datta_cs Jul 15
Excited to welcome @adityagp to @Snowflake and to my team to lead our AI x Data research efforts!

Aditya joins us part-time from the @UCBerkeley where he will continue his faculty role. Aditya and his team have pioneered research on how modern AI should change data systems — from agentic optimization of unstructured data processing (aka the widely used DocETL project), to rethinking database architectures for a world where agents, not humans, are the primary users.

Our relationship with Aditya goes way back — we acquired his company, Ponder, in October 2023, and we're excited to finally welcome him into the fold.

At Snowflake, Aditya will lead our research agenda in this space and work closely with product teams to bring these ideas to life across our AI Data Cloud.

For more on Aditya's research, check out his webpage.

people.eecs.berkeley.edu/~adityagp/
View quoted post on X
4 replies 2 reposts 60 likes 5.8K views View on X