@divytweet

Postdoc @Berkeley_EECS | Prev. PhD @UTCompSci, CSE @iitdelhi

 Berkeley, CA
divyanshusaxena.github.io
Joined June 2017
  • Tweets333
  • Following416
  • Followers262
16Media
Some overdue career updates 🎓 This summer, I defended my Ph.D. after six golden years at UT Austin, working with exemplary supervisors and peers at the @utnslab and @ldosexpedition labs. My thesis builds abstractions and interfaces to make systems programmable, flexible, and robust. Next, I have started in the NetSys lab at @UCBerkeley as a postdoctoral scholar. At Berkeley, I’ll expand my research at the intersection of systems and agents: how do we design systems that remain performant, trustworthy, and maintainable as autonomous agents are increasingly used to build and deploy them? Excited for what lies ahead! Stay tuned!! 🚀
🤖 Made with AI
5
1
45
3,568
**Performance robustness** matters for business-critical applications as much as safety and correctness -- yet, even learned controllers miss latency targets under changing environments. Next week, I will be at NSDI to present my work that addresses this challenge!
1
1
14
543
more replies
This is joint work w/ Gaurav, Jiaxin, @bobowang2014, @IsilDillig, Sanjay, and @adityaakella Catch this work at NSDI first day, first session, first talk (track II)!
1
1
192
And a snap of me presenting this work at NSDI 2026 in Seattle last week.
50
Learned congestion controllers may not always result in good performance - even SOTA learned controllers can result in bandwidth starvation, and other types of poor behavior! Our work on making learned controllers meet desired properties will appear at EuroSys this week!
1
3
26
1,462
Key idea in Canopy: we construct a quantitative feedback from formal verifiers and integrate this feedback *into the training loop* of learned controllers. Thus, the neural network learns to not just optimize for empirical performance but also: how to meet desired properties!
1
2
132
This is joint work with @chenxiyang_ut, Rohit, @kshiteejmahajan, @swarat, and @adityaakella If you are at EuroSys, please check out the talk during Day 2 afternoon session (track B) and reach out if there are any thoughts/questions!
1
163
Divyanshu Saxena retweeted
Aditya Akella
@adityaakellaFeb 25
Our @ACMSIGOPS blog post lays out the LDOS roadmap for learned OS policies. Would love to hear thoughts from the systems + ML community! Co-written with @divytweet Aneesh Durg, Sujay Yadalam, Jiayi (Jane) Chen, Rohit Dwivedula, and Chris Rossbach. Many thanks to the @NSF CISE Expeditions program for supporting this work.
Tianyin Xu
@tianyin_xuFeb 21
LDOS aims to build an intelligent, self-adaptive OS that uses learned policies to share resources efficiently and effectively. It presents a concrete path of using AI to rethink and redesign OS. Divyanshu @divytweet and the @UTCompSci team, led by Prof. Akella @adityaakella, wrote a great article to share their vision, roadmap, and many exciting research directions they've made great progress on. You can find more information at ldos.utexas.edu and @ldosexpedition.
1
4
12
1,643
Thanks for featuring our article! This blog post highlights the motivation behind and the vision for LDOS @ldosexpedition. We are actively working towards the next horizon of intelligent Operating Systems!
Tianyin Xu
@tianyin_xuFeb 21
LDOS aims to build an intelligent, self-adaptive OS that uses learned policies to share resources efficiently and effectively. It presents a concrete path of using AI to rethink and redesign OS. Divyanshu @divytweet and the @UTCompSci team, led by Prof. Akella @adityaakella, wrote a great article to share their vision, roadmap, and many exciting research directions they've made great progress on. You can find more information at ldos.utexas.edu and @ldosexpedition.
1
4
816
Divyanshu Saxena retweeted
ACM SIGOPS@ACMSIGOPSFeb 21
New SIGOPS Blog -- "LDOS: Toward A Learning-Directed Operating System" by by Divyanshu Saxena, Aneesh Durg, Sujay Yadalam, Jane Chen, Rohit Dwivedula, Chris Rossbach, and Aditya Akella. sigops.org/2026/ldos-toward-…
2
1
13
4,258
If you are attending @NeurIPSConf, do check out our poster at the ML for Systems workshop on a novel approach for improving interdependent systems policies by developing shared representations!
1
2
87
We look at this "joint learning" problem in the context of hardware caching and prefetching, and propose two techniques: either learning shared embeddings using a joint encoder, or using contrastive learning to align individual embeddings.
1
65
For decades, OSes have relied on hand-tuned heuristics. What if they could write their own? 🔥 Meet PolicySmith, our LLM-driven framework that generates high-performing, interpretable system code. 📄arxiv.org/abs/2510.08803 at #HotNets25 #LDOS #MLforSystems #CodeGeneration
4
2
8
2,355
Our vision for using LLM-driven search for discovering better systems heuristics is set to appear at HotNets'25! Our thesis is that we can achieve **instance-optimality** using advanced code-generation and reasoning capabilities of LLMs!
LDOS Expedition@ldosexpedition17 Oct 2025
For decades, OSes have relied on hand-tuned heuristics. What if they could write their own? 🔥 Meet PolicySmith, our LLM-driven framework that generates high-performing, interpretable system code. 📄arxiv.org/abs/2510.08803 at #HotNets25 #LDOS #MLforSystems #CodeGeneration
1
2
118
more replies
Crucially, we identify what this high-level specification should contain - abstracting the overall search into a small number of components. We apply this to Web Caching and Congestion Control; and discover heuristics that outperform established baselines in our prototype!
1
2
66
Joint work by Rohit, myself, and our advisors @adityaakella, @daehyeok_kim, and @swarat.
2
85
Load more