Muhammed Ustaömeroğlu (@m_ustaomeroglu) header
Muhammed Ustaömeroğlu (@m_ustaomeroglu) profile picture

PhD student @CarnegieMellon, doing AI and stuff | Ex @Nvidia research intern | UG @Bilkentuniv, EEE & Physics | High school TFL

Joined August 2022
Photo by Muhammed Ustaömeroğlu (@m_ustaomeroglu)Photo by Muhammed Ustaömeroğlu (@m_ustaomeroglu)Photo by Muhammed Ustaömeroğlu (@m_ustaomeroglu)Photo by Muhammed Ustaömeroğlu (@m_ustaomeroglu)Photo by Muhammed Ustaömeroğlu (@m_ustaomeroglu)Photo by Muhammed Ustaömeroğlu (@m_ustaomeroglu)Photo by Muhammed Ustaömeroğlu (@m_ustaomeroglu)Photo by Muhammed Ustaömeroğlu (@m_ustaomeroglu)
I gave Claude our paper (arxiv.org/abs/2605.09294) and asked for an explainer video. It read the paper, wrote the script, animated everything, and narrated it. 6 minutes, 4K. 🤯 @claudeai #Interpretability
2
138
At least it Claude is honest
2
108
Muhammed Ustaömeroğlu retweeted
At this rate we'll probably see GPT-6 before GTA 6
106
269
110
2,291
590,434
Muhammed Ustaömeroğlu retweeted
This is crazy. The incredible Yitan Zhang (newyorker.com/magazine/2015/…) worked on proving this conjecture for 7 years. Moh, his advisor, wrote that Zhang "failed miserably" in proving the Jacobian conjecture, "never published any paper on algebraic geometry" after leaving Purdue, and "wasted seven years of his own life and my time". What a twist.
hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final ((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0)
31
163
31
2,224
535,375
I have 4 papers at #ICML2026 in Seoul this week, spanning the main track and workshops. I’ll present BLOCK-EM and RET; my coauthors will present the other two. Times below are KST. If any interest you, stop by a poster or DM me for coffee.
1
4
161
PubSwap: Public-Data Coordination for Federated RLVR Public-data off-policy steps improve federated RLVR coordination and efficiency. With Anupam Nayak, Baris Askin, Carlee Joe-Wong & Gauri Joshi. RLxF: Fri Jul 10 | DEMO: Sat Jul 11 arxiv.org/abs/2604.12160
34
Muhammed Ustaömeroğlu retweeted
Me using Claude Opus 4.8 to rename a file
1,697
9,046
1,499
73,938
44,435,221
Excited to share our new paper: “Towards Effective Theory of LLMs: A Representation Learning Approach”, with my advisor @guannanqu. The figure is from project webpage is probably the best way to quickly get an intuition for the framework: ustaomeroglu.github.io/RET/
1
2
171
In the paper, we show that RET can: * reveal temporally consistent “mental-state” trajectories during reasoning, * capture high-level semantic structure, * predict behavioral outcomes such as sycophancy early, * and provide causal handles for steering generations.
1
34
8/10 We also analyze the hidden-state subspace linked to emergent misalignment. It is not highly low-dimensional: with prolonged fine-tuning, misalignment can partly re-emerge via alternative pathways, though broader blocking mitigates this.
1
32
9/10 I’m grateful to my advisor @guannanqu for his guidance and support throughout this project. Looking forward to presenting this work at ICML 2026 in Korea. #ICML2026 #MachineLearning #AIAlignment #AISafety #LLMs #Interpretability #MechanisticInterpretability
30