Niloofar
@niloofar_mire
Followers
8K
Following
78K
Media
212
Statuses
4K
Niloofar Mireshghallah — incoming asst. prof @LTIatCMU @CMU_EPP, RS in @AIatMeta, postdoc @uwcse, Ph.D. @ucsd_cse, former @MSFTResearch -Privacy, ML, NLP
San Francisco, CA
Joined May 2013
📣Thrilled to announce I’ll join Carnegie Mellon University (@CMU_EPP & @LTIatCMU) as an Assistant Professor starting Fall 2026! Until then, I’ll be a Research Scientist at @AIatMeta FAIR in SF, working with @kamalikac’s amazing team on privacy, security, and reasoning in LLMs!
226
67
1K
I'm recruiting PhD students! I'm interested in: 1. Understanding how LLMs 'see' the world (ex: LMs can't see conspicious omissions, see AbsenceBench) 2. How can we make things with LLMs that have never been made before? (ex: Communnication Games, see 📌) 3. See my other posts :)
0
3
34
All human beings are spiritual in some way. Some realized it and others don't. The inner soul of every human is searching for a connection with super. Just like neurons search for bonding in brain. By nature, it is made difficult to find who we are. We are surviving until we find
0
0
18
There are many anecdotal cases of reward hacking in LLMs, but we can now systematically induce and measure this “rogue” behavior (almost) in-the-wild by creating deliberate conflicts between the natural-language specification and the test cases. Models take shortcuts, often
New research with @AdtRaghunathan, Nicholas Carlini and Anthropic! We built ImpossibleBench to measure reward hacking in LLM coding agents 🤖, by making benchmark tasks impossible and seeing whether models game tests or follow specs. (1/9)
0
2
43
Harder, better, ... faster?😉 Come join us at Databricks HQ on Thursday November 6th for the launch of Terminal-Bench 2.0 and a secret new project. Luma link below 🔗
1
0
16
Professor Lorrie Cranor (@lorrietweet), @CyLab Director, is one of five recipients of the 2025 @acrmuseum's Stibitz-Wilson Awards, recognizing pioneering individuals whose work has powerfully impacted modern life.
cylab.cmu.edu
Lorrie Cranor, Director of the CyLab Security and Privacy Institute, is one of five recipients of the 2025 American Computer & Robotics Museum’s (ACRM) Stibitz-Wilson Awards, recognizing pioneering...
0
1
3
in case you missed recursive language models (or if you have better task suggestions for us, please!!!!)
Very cool work by @a1zhang and @lateinteraction Instead of calling an LM to solve a problem, let it be able to agentically call an LM that works over an environment, storing prompt and context (that evolve over time) The root LM then answers using all info aggregated by the
7
8
155
I will be giving a talk at @ETH_AI_Center next week, on RLVR for verifiable instruction following, generalization, and reasoning! 📢 Join if you are in Zurich and interested in hearing about IFBench and our latest Olmo and Tülu works at @allen_ai
0
6
68
오랜만에 서울 왔습니다. 언제나처럼 학생분들과 얘기하고 싶어요. last-minute이긴 한데 오늘 저녁에 맥주 한 잔 하며 저와 같이 대화 나누시고 싶은 학생분들이 있으려나 모르겠네요. 만약 관심 있으시며 rsvp, please! (rsvp link below)
1
8
60
Hubble is finally out! We used 200k GPU hours from @NSF NAIRR and @nvidia to build a comprehensive resource for the scientific study of LLM memorization. Fully open-source models & data up to 8B params + 500B tokens with controlled data insertion to study memorization risks 🔭✨
Announcing 🔭✨Hubble, a suite of open-source LLMs to advance the study of memorization! Pretrained models up to 8B params, with controlled insertion of texts (e.g., book passages, biographies, test sets, and more!) designed to emulate key memorization risks 🧵
2
10
53
We’re doing something big in the heart of NYC to highlight an even bigger issue: how exposed our lives can be online. Stay tuned. 🤫
2
4
16
Congratulations to @ZhiruoW for an honor well deserved!
🎉Thrilled to be named a Google PhD Fellow! Feeling incredibly grateful for my advisors, collaborators, and friends who’ve supported and inspired me along the way 🫶 Looking forward to the journey ahead!
1
1
15
If writing technical blogs/tutorials on AI (decentralized / federated / privacy-preserving / etc.) that get on #HackerNews / #Reddit / X sounds like a fun day job... DM me. (I would mentor you.)
4
6
39
Check out our new paper BehaviorSFT! In clinical settings, different scenarios require vastly different reactive--proactive actions by the agent🤖 We introduce a **proactiveness token** to first predict the action type, then condition on this to take the appropriate action👇
Can agents calibrate its own behavior in different contexts? Our BehaviorSFT paper introduces a benchmark and training strategy with behavior tokens that makes agent to adapt proactiveness, learning when to stay reactive and when to speak up. Arxiv: https://t.co/yf3724unaY
0
5
46
Congratulations to CyLab's 2025 Presidential Fellows! Each year, CyLab recognizes high-achieving Ph.D. students pursuing security and/or privacy-related research with a CyLab Presidential Fellowship that covers one year of tuition.
cylab.cmu.edu
Each year, CyLab recognizes high-achieving Ph.D. students pursuing security and/or privacy-related research, with a CyLab Presidential Fellowship, covering an entire year of tuition.
0
1
7
NYU is recruiting faculty fellows! Happy to chat with anyone who is considering this as an option:
0
14
58
If you or someone you know is interested in working on Multi-Agent AI as a postdoc with my lab, please let me know! I'd be excited to support applications for this fellowship:
computing.mit.edu
2
14
35
Has the week been unFAIR? I’m sorry to hear it! Did you know Microsoft is hiring in genAI across divisions? Check out their portal for a rich set of listings. Here are some examples: https://t.co/RGeLnr8lS9
https://t.co/FgTQUdmcBp
0
13
48
The concept of dual citizenship for Christians—residing in the world while holding citizenship in heaven—offers a profound foundation-for living. As Scripture declares "Our citizenship is in heaven, and from it we await a Savior, the Lord Jesus Christ." This spiritual identity
0
0
4
Anyone who was affected by the FAIR layoffs and is interested in foundational research to understand model capabilities, please dm me! We're <30 people, well-founded, and you'll work on exceptionally impactful research with no politics
@xianjun_agi @johnschulman2 @METR_Evals is hiring! Shoot me a DM if you're interested. We try to empirically measure the capabilities and risks of LLMs. Here's an example of our work. https://t.co/4WBLvVrSuL
5
11
137
Reach out if you were impacted by the Meta layoffs and want to work on frontier research!
0
6
99
UTCS is hiring in all areas, including PL! Please DM me if you are on the job market this year and interested in joining our wonderful department :)
1
27
79
If you are at Johns Hopkins tomorrow stop by --- also on Zoom:
hub.jhu.edu
Aaron Roth, a professor of computer and cognitive science in the Department of Computer and Information Science at the University of Pennsylvania, will give a talk titled "Agreement and Alignment for...
0
5
22
Thanks @thinkymachines for supporting Tinker access for our CS329x students on Homework 2 😉
Its not even been a month since @thinkymachines released Tinker & Stanford already has an assignment on it
6
32
580
So, if illegal drugs are such a threat to our sovereignty that our Wacko Killer is randomly blowing up boats on the open sea, threatening Venezuela with military action, why not cartel-infested Mexico, perhaps a couple A-10s shooting up drug-laden vehicles on the highways of
4
1
22