jubaz.bsky.social
Juba Ziani
@jubaz.bsky.social
620 followers 450 following 81 posts
Assistant professor at Georgia Tech in ISyE. I do mechanism design, differential privacy, fairness, and learning theory, mostly. Postdoc @Penn; Ph.D. @Caltech; MSc @Columbia and @Supélec. He/him.
Posts Media Videos Starter Packs
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · 19d
Hello, this is now accepted at Neurips 2025! Come check out our poster in December :)
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Aug 11
Check out our @let-all.com blog post on strategic classification and around recent work with Valia and @charapod.bsky.social!
let-all.com
let-all.com @let-all.com · Aug 11
New blog post on the Learning Theory Alliance blog, by Diptangshu Sen (@dsensei.bsky.social) and Juba Ziani (@jubaz.bsky.social), on strategic classification. Click through to read more!

www.let-all.com/blog/2025/08...
1 3
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · 19d
Thanks to Yunzong Xu and Bhaskar Ray Chaudhuri for the invite!
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · 19d
Can't recommend the Allerton conference at UIUC enough! Went for the first time this year and really enjoyed the smaller scale, being able to talk to so many great researchers, the community, and the great talks! Definitely got quite a bit out of that and a million new ideas
1 4
Reposted by Juba Ziani
ccanonne.github.io
Clément Canonne @ccanonne.github.io · Aug 25
New differential #privacy textbook in town: "DP in Artificial Intelligence: From Theory to Practice", by @nandofioretto.bsky.social and @vanhentenryck.bsky.social. Open access, w/ chapters by @jubaz.bsky.social, @grahamrc.bsky.social, and @stein.ke!

www.nowpublishers.com/article/Book...
Front cover: Differential Privacy in Artificial Intelligence: From Theory to Practice, now Publishers
1 7 15
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Aug 22
It's also the tone of the PC. They keep changing the process mid-day, shortening the timeline for ACs, and instead of acknowledging this, blaming the ACs and sending repeatedly threatening emails for ha being one day late on a task that we had a half the original assigned time for
2
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Aug 11
Check out our @let-all.com blog post on strategic classification and around recent work with Valia and @charapod.bsky.social!
let-all.com
let-all.com @let-all.com · Aug 11
New blog post on the Learning Theory Alliance blog, by Diptangshu Sen (@dsensei.bsky.social) and Juba Ziani (@jubaz.bsky.social), on strategic classification. Click through to read more!

www.let-all.com/blog/2025/08...
2 4
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Jul 19
The tl;dr is that balanced/fair algorithms do not necessarily comes at a cost. When you take incentives around data and network effects into account, fairness (here, through representativeness) can come at no cost.

And the paper that started it all: arxiv.org/abs/2501.19294
NSF Award Search: Award # 2504990
Collaborative Research: III: Medium: Incentives and interventions for robust networked data exchange
www.nsf.gov
1 6
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Jul 19
Excited to announce that I just got another NSF grant this week! This collaborative with Prof. Augustin Chaintreau at Columbia University.

More info here: www.nsf.gov/awardsearch/...
NSF Award Search: Award # 2504990
Collaborative Research: III: Medium: Incentives and interventions for robust networked data exchange
www.nsf.gov
4 19
Reposted by Juba Ziani
ccanonne.github.io
Clément Canonne @ccanonne.github.io · Jul 8
Pretty stoked to announce that I won't be presenting my paper at #FOCS2025: but one of my talented coauthors will have to come here 🌏 to do so!

"Instance-Optimal Uniformity Testing and Tracking," with Guy Blanc (Stanford) and Erik Waingarten (UPenn). (arXiv coming soon!)
1 1 24
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Jun 13
The cointreau or the apricot? I found that every time I put cointreau in a cocktail, I should have used half.
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Jun 13
How did it turn out
1
Reposted by Juba Ziani
kvachai.bsky.social
Krishna Acharya @kvachai.bsky.social · Jun 9
1/8 Happy to share that our paper GLoSS: Generative Language Models with Semantic Search for
Sequential Recommendation is accepted at the KDD OARS workshop! 🎉
Paper, code: github.com/krishnachary...
This is joint work with my wonderful collaborators
@apetrov.bsky.social and @jubaz.bsky.social !
GitHub - krishnacharya/GLoSS: GLoSS: Generative Language Models with Semantic Search for Sequential Recommendation
GLoSS: Generative Language Models with Semantic Search for Sequential Recommendation - krishnacharya/GLoSS
github.com
1 1 1
Reposted by Juba Ziani
propublica.org
ProPublica @propublica.org · Apr 11
THREAD: Under a new law, thousands of prisoners in Louisiana have been cut off from ever getting a chance at parole.

Why?

Because an algorithm said so. 1/
490 4.7K 11K
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · May 1
Hello, I'm off Twitter forever! This is the main place to find me now :)
10
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Apr 7
Hi everyone!

First, wanted to let you all know that the amazing @charapod.bsky.social is on Bluesky! Please make sure to follow her :)

Second, she wrote a really cool survey about the state of strategic classification that you should definitely read: www.sigecom.org/exchanges/vo...
www.sigecom.org
11
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
With this work, we aim to add nuance to the discourse that decentralization on its own may not be the solution---rather, centralized decision-making should be more fine-grained to go beyond naive metrics, and understanding diversity of backgrounds/signals that make up qualified candidates.
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
This creates unfairness where *equally qualified* candidates are treated disparately based on, for example, how recognizable their alma matter is.
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
The easy way out here is spectacularly bad: hire candidates that they fully understand/have more information about, rather than candidates that are riskier (i.e., a top student from a small, not well-known high school).
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
Our main insight is as follows: while decentralization is useful to score candidates beyond just generic characteristics/take into account how successful they are expected to be for a specific team, our mathematical model shows that such decentralized designer will always take the easy way out.
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
In stage 2, decentralized entities (professors in a university or specific teams in a company) make admissions/hiring decisions based not only on "quality" (again, which is imperfectly perceived), but also how their specific attributes contribute to the specific team they aim to join.
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
We propose a mathematical model of 2-stage admission/hiring that questions this conventional wisdom.

In stage 1, a centralized designer makes initial hiring/admissions decisions based solely on a candidate's perceived "quality" (a noisy signal of their competences).
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
In particular, applicants that look atypical may be ignored (e.g., top and highly qualified applicants from lesser known schools, or applicants with less access to extracurricular activities to add to their CV).
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
The classical wisdom of the crowd is that centralization is bad: centralization tends to rely on generic decision-making rules and AI tools that optimize for historical admissions criteria without considering the diversity of applicant backgrounds.
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
Here, we study the impact of centralization vs decentralization in hiring and admissions.
1
jubaz.bsky.social
Juba Ziani @jubaz.bsky.social · Mar 10
A last work I wanted to talk about (I promise I will stop spamming after this) with my student Diptangshu Sen and collaborator Ben Fish is here: www.arxiv.org/abs/2502.07792
Centralization vs Decentralization in Hiring and Admissions
There is a range of ways to organize hiring and admissions in higher education, as in many domains, ranging from very centralized processes where a single person makes final decisions to very decentra...
www.arxiv.org
1 6