Narmeen Fatimah Oozeer

Joined

January 2025

Names

Narmeen Fatimah Oozeer (Preferred)

Narmeen Oozeer

Emails

****@gmail.com (Confirmed)

****@withmartian.com (Confirmed)

****@uottawa.ca (Confirmed)

Personal Links

Google Scholar

DBLP

Semantic Scholar

Career & Education History

Researcher

Research, Martian Learning (withmartian.com)

2024 – 2025

Undergrad student

Mathematics, University of Ottawa (uottawa.ca)

2020 – 2024

Advisors, Relations & Conflicts

Coauthor

Alice Rigg

2024 – 2025

Expertise

Reinforcement Learning

2024 – 2025

Linear Algebra

2024 – 2025

Mechanistic Interpretability & Alignment

2023 – 2025

Publications

The Capability Frontier: Benchmarks Miss 82% of Model Performance
Bradley Fowler, Ryan Smith, Daniel Thi Graviet, William Myers, Joshua Greaves, Narmeen Fatimah Oozeer, Antía García, Philip Quirke, Fazl Barez, Amir Abdullah, Shriyash Kaustubh Upadhyay
- ICLR 2026 Workshop DATA-FM Oral
- Readers: Everyone
The Capability Frontier: Benchmarks Miss 82% of Model Performance
Bradley Fowler, Ryan Smith, Daniel Thi Graviet, William Myers, Joshua Greaves, Narmeen Fatimah Oozeer, Antía García, Philip Quirke, Fazl Barez, Shriyash Kaustubh Upadhyay
- ICLR 2026 Workshop LLM Reasoning
- Readers: Everyone
Are LLM Evaluators Really Narcissists? Sanity Checking Self-Preference Evaluations
Dani Roytburg, Matthew Bozoukov, Matthew Nguyen, Jou Barzdukas, Mackenzie Puig-Hall, Narmeen Fatimah Oozeer
- ICML 2026 regular
- Readers: Everyone
Breaking the Mirror: Examining Self-Preference in LLM Evaluators through Activation-Based Representations
Dani Roytburg, Matthew Bozoukov, Hongyu Fu, Matthew Nguyen, Jou Barzdukas, Narmeen Fatimah Oozeer
- NeurIPS 2025 LLM Evaluation Workshop Poster
- Readers: Everyone
Approximating Human Preferences Using a Multi-Judge Learned System
Fernando Avalos, Eitan Sprejer, Augusto Mariano Bernardi, José Pedro Brito de Azevedo Faustino, Jacob Haimes, Narmeen Fatimah Oozeer
- NeurIPS 2025 - Reliable ML Workshop
- Readers: Everyone
Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators
Dani Roytburg, Matthew Nguyen, Matthew Bozoukov, Hongyu Fu, Jou Barzdukas, Narmeen Fatimah Oozeer
- NeurIPS 2025 - Reliable ML Workshop
- Readers: Everyone
Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators
Dani Roytburg, Matthew Nguyen, Matthew Bozoukov, Jou Barzdukas, Hongyu Fu, Narmeen Fatimah Oozeer
- Mech Interp Workshop (NeurIPS 2025) Poster
- Readers: Everyone
Bilinear Convolution Decomposition for Causal RL Interpretability
Sinem Erisken, Alice Rigg, Narmeen Fatimah Oozeer
- Mech Interp Workshop (NeurIPS 2025) Poster
- Readers: Everyone
Position: Require Frontier AI Labs To Release Small "Analog" Models
Shriyash Kaustubh Upadhyay, Philip Quirke, Narmeen Fatimah Oozeer, Chaithanya Bandi
- NeurIPS 2025 Position Paper Track
- Readers: Everyone
Beyond Monoliths: Expert Orchestration for More Capable, Democratic, and Safe Large Language Models
Philip Quirke, Narmeen Fatimah Oozeer, Amir Abdullah, Jason Hoelscher-Obermaier, Clement Neo, Jeff M. Phillips, Joshua Greaves, Chaithanya Bandi, Fazl Barez, Shriyash Kaustubh Upadhyay
- Submitted to NeurIPS 2025 Position Paper Track
- Readers: Everyone

View all 13 publications

Co-Authors

View all 33 co-authors