Co-Founder @ Coactive AI
San Francisco, CA
Joined June 2022
- Tweets51
- Following115
- Followers117
- Likes200
Open benchmarking done right. @nvidia's Nemotron Nano 12B hitting 30B-level quality significantly lower cost for workloads. Proud to have them as a MediaPerf governance board partner 🚀
Our Nemotron Nano 12B v2 VL brings video understanding on-prem.
MediaPerf benchmark launched by Coactive ranks our 12B model on par with 30B-size models at less than half the footprint:
✅ Cost Efficiency: Lowest cost for Tagging Refinement workload.
✅ Pro-Grade Quality: 0.299 F1 on real-world media tasks.
✅ Massive Throughput: 4.48 hrs video/hr - 15% faster than the leading 30B OS alternative.
✅ Sovereignty: Self-hostable, open model for every developer worldwide.
✅ Transparency: Open training datasets, techniques, and libraries.
🔗 mediaperf.org/
Will Gaviria Rojas retweeted
Our Nemotron Nano 12B v2 VL brings video understanding on-prem.
MediaPerf benchmark launched by Coactive ranks our 12B model on par with 30B-size models at less than half the footprint:
✅ Cost Efficiency: Lowest cost for Tagging Refinement workload.
✅ Pro-Grade Quality: 0.299 F1 on real-world media tasks.
✅ Massive Throughput: 4.48 hrs video/hr - 15% faster than the leading 30B OS alternative.
✅ Sovereignty: Self-hostable, open model for every developer worldwide.
✅ Transparency: Open training datasets, techniques, and libraries.
🔗 mediaperf.org/
Today we’re launching MediaPerf – the first open AI benchmark built specifically for the media industry. Quality + cost + latency reported together. Governance board: AWS, Business Insider, Harvard, Georgia Tech, NBCUniversal, NVIDIA.
See more at mediaperf.org
Will Gaviria Rojas retweeted
.@SFResearch’s new series “AI Research Lab - Explained” just dropped!
First up? See how we fine-tune specialized models to predict actions, not just language—enabling faster, more precise execution of real-world tasks.
⏯️ Watch and subscribe on YouTube: youtube.com/watch?v=vlvv4Zeh…
🎬 NOW LIVE: "The AI Research Lab - Explained" debuts with our groundbreaking work on Large Action Models!
Watch now: bit.ly/4kfipp4
Watch as Shelby Heinecke @shelbyh_ai reveals how we're training these specialized models to generate precise, executable actions rather than merely text—enabling AI agents to respond in less than a second.
#LAMs #EnterpriseAI
Will Gaviria Rojas retweeted
1/ Why has Autonomous Driving (AD) failed? We believe that the key to unlocking AD may be in 1) collecting rich out-of-distribution data from emerging economies such as #Peru; and 2) developing brain-aligned computer vision models #NeuroAI
artificio.org/blog/why-has-a…
Will Gaviria Rojas retweeted
a report summarizing topics and events discussed at the inaugural DMLR workshop @icmlconf this year is now available at
arxiv.org/abs/2311.13028
consider joining the community discord at discord.gg/FswYXMv4j9 for contributions to ongoing and future activities
Will Gaviria Rojas retweeted
Behind the #1 AI CRM are scientists, researchers, and engineers enhancing @Salesforce products and pushing the field of AI forward.
@silviocinguetta, @shelbyh_ai, & @nikhil_ai share how @SFResearch has influenced CRM & society.
Get the full story: salesforce.com/news/stories/… #DF23
Will Gaviria Rojas retweeted
Learn about our Salesforce, in-house large foundation models for text and multi-modal applications. Discover how our trusted #AI models can transform your business use cases.
Join us 9/14 at 9 a.m. in AI Landing! #DF23 @JotyShafiq @shelbyh_ai @shugerdou
reg.salesforce.com/flow/plus…
Will Gaviria Rojas retweeted
@willg_ai, co-founder of @CoactiveAI, is live on stage 2 of the LLM in Production conference! Bringing structure to unstructured data. Join us and throw your questions and thoughts in the chat!
Will Gaviria Rojas retweeted
Our #DMLR workshop at @icmlconf accepted 108 submissions!
Thanks to all the authors and reviewers for your hard work. We look forward to seeing you in Hawaii! 🏝️
ALT A slide for the Data-centric Machine Learning Research (DMLR) workshop with a fantastic line of speakers, including Andrew Ng, Dina Machuve, Isabelle Guyon, Masashi Sugiyama, Mihaela van der Schaar, Olga Russakovsky, and Peter Mattson. A bar chart shows 33 accepted extended abstracts and 75 research papers.
Excited to share that I'll be giving a lightning talk tomorrow at the LLMs in Production conference
I'll cover key lessons we've learned at @CoactiveAI in designing #ai systems to process unstructured data at scale
A big shoutout to @Dpbrinkm and the @mlopscommunity team!🤘🙌
For folks in #AI working with distributed computing + complex data types (images, audio, etc), @daft_dataframe is a game changer.
Fast, performant and pythonic 👌
@Sammy_Sidhu and Jay are truly ahead of the curve
🚀 Introducing Daft: A High-Performance Distributed Dataframe Library for Multimodal Data 📊🔥
blog.getdaft.io/p/introducin…
🧵 1/5
Will Gaviria Rojas retweeted
🚀 Introducing Daft: A High-Performance Distributed Dataframe Library for Multimodal Data 📊🔥
blog.getdaft.io/p/introducin…
🧵 1/5
Why do we benchmark ML but not the data it relies on?
To close this gap, we are excited to announce #DataPerf - a collaboration across academia and industry to benchmark data-centric AI.
Learn more about the image classification benchmark @codyaustun and I created below:
Announcing DataPerf 2023 challenges submission extension to July 1st! Winners will be invited to present at ICML23 & submit for publication to the Data for ML Research journal. Join the challenges today!
mlcommons.org/en/news/datape…
#DataPerf #MLCommons #startups #academia
Will Gaviria Rojas retweeted
Only one week is left to submit your paper to our data-centric ML research (DMLR) workshop at @icmlconf 2023 hosted in Hawaii!🏝️
We have a fantastic line of speakers, including @AndrewYNg, @orussakovsky, @DMachuve, @MihaelaVDS, and more! (1/4)
ALT A CFP slide for the Data-centric Machine Learning Research (DMLR) workshop at ICML 2023 with a fantastic line of speakers, including Andrew Ng, Dina Machuve, Isabelle Guyon, Masashi Sugiyama, Mihaela van der Schaar, Olga Russakovsky, and Peter Mattson. The submission deadlin is May 26, 2023.
Very excited to be speaking at NorCalDB Day about Bringing Structure to Unstructured Data with an AI-First System Design! If you're also in attendance, tune in at 3:45pm PT.
microsoft.com/en-us/research…
#NorCalDBDay #datacentricAI
Will Gaviria Rojas retweeted
How teachers made a difference in the trajectory of one kid’s life
Learn more about @codyaustun in my latest @educationweek column “Ask a Psychologist.” edw.link/thrivenowcody #helpstudentsthrive
Will Gaviria Rojas retweeted
This week @codyaustun & @willg_ai co-Founders @CoactiveAI shared their collective story
Meeting as CS freshmen @MIT + reconnecting after their PhDs to build @CoactiveAI, the HUGE market oppty, why unstructured visual data is hard, and the tech that makes @CoactiveAI so special!
Will Gaviria Rojas retweeted
Enterprises are generating an explosion of visual content, yet this kind of content is the most challenging form of information to analyze as it is typically unstructured.
@CoactiveAI brings structure to visual content through machine learning. Industries with high volume of visuals include retail, social media, medical imaging, gaming, and autonomous vehicles.
👀🔥 New piece!
"What to Watch in AI" highlights 10 promising companies transforming industries with AI.
Jump into the full piece or keep scrolling for the tweet-sized summaries of each.
thegeneralist.substack.com/p…
Will Gaviria Rojas retweeted
@ExceptionalCap is excited to be a part of @Coactive AI’s journey as they help the world finally structure its unstructured visual data and bring #contexttocontent! Congratulations, @codyaustun and @willg_ai #AI #MachineLearning #datacentricai
Structuring unstructured image and video data opens up exciting new possibilities for data analysis. Coactive AI makes it possible for any organization to do so. We are excited to be a part of the journey, congratulations to @CoactiveAI @a16z @BessemerVP businessinsider.com/coactive…