Nazneen Rajani

Also known as Nazneen Fatema Rajani

Nazneen Rajani is an AI researcher whose work links explainable models, language-model alignment and evaluation. She coauthored a neural-network tic-tac-toe paper in 2011, completed a computer science PhD at UT Austin in 2018, and coauthored ERASER, a benchmark for NLP explanations, in 2019. At Hugging Face, she coauthored the StackLLaMA fine-tuning guide in April 2023; her Zephyr report explored language-model alignment with AI feedback in October 2023. She joined the UN Secretary-General’s High-Level Advisory Body on AI in October 2023, and MIT Technology Review named her to its 2024 Innovators Under 35 list. As of October 2026, she is cofounder and CEO of Collinear AI, which develops training data, evaluation tools and realistic environments for frontier AI models.

Now at: Collinear AI
Works: 83
Citations: 3,818
h-index: 24
Badges: Founder
Bridge

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

2011

A neural-network tic-tac-toe paper

In 2011, she coauthors a paper that applies a neural network and a Hamming-distance method to tic-tac-toe.
2018

A computer science PhD

She completes a computer science PhD at The University of Texas at Austin in 2018; her thesis is titled “Explainable Improved Ensembling for Natural Language and Vision.”
2019

The ERASER benchmark

In 2019, she coauthors ERASER with Salesforce Research collaborators, a benchmark that collects human-marked rationales for NLP tasks and measures their alignment with model explanations and influence on predictions.
April 2023

The StackLLaMA guide

In April 2023, she coauthors Hugging Face’s StackLLaMA guide, which fine-tunes LLaMA using supervised fine-tuning, reward modeling and reinforcement learning from human feedback.
October 2023

The Zephyr report

In October 2023, she coauthors the Zephyr report, which distills user-intent alignment into a 7B language model using preference data from AI feedback.
October 2023

The UN AI advisory body

In October 2023, she joins the UN Secretary-General’s High-Level Advisory Body on AI as a lead researcher at Hugging Face. The body brings experts together to build scientific consensus on AI risks and strengthen international cooperation on AI governance.
2024

Innovators Under 35

MIT Technology Review names her to its 2024 Innovators Under 35 list for work on AI systems that could make models safer and more reliable.
As of October 2026

CEO at Collinear AI

As of October 2026, she is cofounder and CEO of Collinear AI, which develops training data, evaluation tools and realistic environments for frontier AI models. Her research spans model evaluation, alignment and agent reliability.