Pratiksha Thaker (@prthaker_) profile picture

Research @databricks, recently postdoc @mldcmu. tweets are my own

Joined October 2024
Photo by Pratiksha Thaker (@prthaker_)Photo by Pratiksha Thaker (@prthaker_)
And a best artifact award as well!? Check out Francisco's demo & experiments here: fchamicapereira.github.io/ag…
Really excited about some new work that @fchamicapereir1 is presenting at HotNets today! (w/ the inimitable @justinesherry) We revive some classic work on the "knowledge plane" updated to a modern, agentic world. What role can agents play in managing the Internet?
2
33
Really excited about some new work that @fchamicapereir1 is presenting at HotNets today! (w/ the inimitable @justinesherry) We revive some classic work on the "knowledge plane" updated to a modern, agentic world. What role can agents play in managing the Internet?
2
1
1
5
204
This might sound like a far-fetched idea, but @fchamicapereir1's experiments show that agents can already be effective in critical settings like preventing BGP attacks. And, naturally, we're excited about the rich set of research problems that arise from thinking about the risks.
1
1
37
Pratiksha Thaker retweeted
Huge thanks to Pratiksha for writing a thread about our works here!! Highly encourage you to check it out. For details on tomorrow’s presentations: 10:30-12:15 - Position: Preventing AI-Generated CSAM Necessitates New Approaches to AI Safety (Spotlight) - Hall A #3013 Co-led this work with Rebecca Portnoff from @thorn, and will be presenting with her colleague and our coauthor Michael Simpson! We outline 15 open problems in AI safety that remain critically unaddressed by current work, and are research barriers to preventing AIG-CSAM. These range from concept fusion and data cleaning to unlearning and robust watermarking. Hope to see you there and spark some great discussions, about what can be done to attack this problem as ML researchers. Michael is highly knowledgeable about the policy and governance landscape surrounding this problem, so encourage anyone interested to come ask those questions too! (Link to all 15 papers in thread). 2:30 - 4:15 - Membership Inference Attacks for Unseen Classes - Hall A #4004 Co-led this work with @prthaker_ who was the driving force behind the initial idea. We’ve been working on this paper almost since I first joined CMU in 2024, and I’m really happy to finally present it here on behalf of our small team! In addition to advancing our understanding of MIAs (and showing we could audit models for CSAM in training data without accessing CSAM), this paper was one of my first ML research experiences and taught me so much about producing high -quality research, putting theory first, and how to persevere on a seemingly impossible research problem - detecting whether a model was trained on data without any in-distribution samples! Come check out the poster, super stoked to share more. Huge thanks to my research advisor @gingsmith whose guidance was so insightful on both papers, and all coauthors involved - @prthaker_, @zstevenwu, @chhaviyadav_, Kevin Kuo - and Rob Wang, Michael Simpson, and Rebecca Portnoff from Thorn! These works were truly cross-domain team efforts. See you all tomorrow!
🌟 I'm not at ICML this year, but thrilled about some of our work that will be presented there! 🧵
1
2
14
2,603
🌟 I'm not at ICML this year, but thrilled about some of our work that will be presented there! 🧵
1
4
1
11
3,853
Check out these works and chat with @neilkale and Michael Simpson from @thorn at ICML!
1
2
149
(It's hard to overstate how excited I am about these papers - the culmination of months to years of hard work with genuinely lovely, thoughtful collaborators who care a lot about addressing real-world problems with deeply technical solutions.)
2
94
Pratiksha Thaker retweeted
FlashOptim is at #ICML2026 this week 🚀🚀 My amazing colleague @jjgort is presenting this Thursday. Drop by to learn more about our work, hear some fun stories behind it, or just to say hi !!
🚀 Today we’re releasing FlashOptim: better implementations of Adam, SGD, etc, that compute the same updates but save tons of memory. You can use it right now via `pip install flashoptim`. 🚀 arxiv.org/abs/2602.23349 A bunch of cool ideas make this possible: [1/n]
9
16
2,537
Pratiksha Thaker retweeted
🇰🇷 Excited to present 2 papers on AI safety at ICML 2026 in Seoul next week, including a first-author spotlight! (TW) Both papers revolve around preventing AI-generated child sexual abuse material. (1) explains how existing AI safety work fails to address the real considerations of working with AIG-CSAM, such as data access restrictions. (2) explores how quantile membership inference attacks (QMIA) can allow us to audit models without ever accessing in-distribution samples, e.g., audit models for CSAM w/o data access. Stay tuned, will be posting more about these papers over the coming days!
1
3
35
2,662
Pratiksha Thaker retweeted
Meet KARL, an RL'd model for document-centric tasks at frontier quality and open source cost/speed. Great for @databricks customers and scientists (77-page tech report!) As usual, this isn't just one model - it's an RL assembly line to churn out models for us and our customers 🧵
9
45
17
243
76,969
Pratiksha Thaker retweeted
[10/n] Paper: papers.ssrn.com/sol3/papers.… Website: aichildsafety.github.io Joint work between @thorn and @mldcmu: @neilkale*, R. Portnoff*, @prthaker_ , M. Simpson, R. Wang, K. Kuo, @chhaviyadav_ , @gingsmith We welcome collaboration. Reach out if you're working on these problems!
1
3
193
I'm so excited that this work is available after a year of carefully curating open problems with our collaborators. It was inspired by real issues we faced applying research techniques to problems in child safety, and we hope this work can help amplify those lessons.
[1/n] Open Problems in AI Child Safety AI is misused to generate CSAM at alarming scale. 400% increase in AI-generated CSAM since 2024 (IWF). 1 in 17 teens are victimized by deepfake nudes. We outline 15 open problems where AI safety research can help. 🔗aichildsafety.github.io
1
6
1,797