About Me

Cristobal Eyzaguirre

Ph.D. student in Stanford's Vision and Learning Lab (SVL), co-advised by Juan Carlos Niebles and Jiajun Wu, studying efficient video understanding, multimodal reasoning, and evaluation of generative video models. Previously I completed an M.Sc.Eng. at Pontificia Universidad Catolica de Chile, where I worked on machine reasoning, meta-learning, and adaptive computation. I have also been a course assistant for CS231n at Stanford twice, in Spring 2025 and Spring 2026. Previously, at Pontificia Universidad Catolica de Chile, I supported courses in artificial intelligence, deep learning, and software engineering while teaching in the institution's AI diploma program. I'm passionate about the outdoors (mountain biking, ultra-running, skiing, hiking, etc.) and scientific innovation in general.

My Career

Stanford University

Ph.D. student in Stanford Vision and Learning Lab (SVL) co-advised by Juan Carlos Niebles and Jiajun Wu.

Fall 2021 - Present
Ph.D. Student

Meta

Worked on multimodal post-training for video understanding in Llama within Meta's GenAI Multimodal team.

Summer 2025
Research Scientist Intern

Toyota Research Institute

Characterizing question complexity in video question answering.

Summer 2023
Research Intern

Google Research

Researched multimodal algorithms for active speaker detection, and integrated the model into a production MediaPipe pipeline (included face detection, facial landmark extraction, and audio processing models).

Oct. 2020 - Jan. 2021
Research Intern

Zippedi

Research and implementation of computer vision algorithms for the automatic recognition of products in store shelves.
Models run locally on embedded devices, or on cloud GPU instances.

Winter 2019
Research Intern

Pontificia Universidad Católica de Chile

Worked on Adaptive Computation Time on recurrent and non-recurrent models (and ensembles).
Graduated with highest distinction.

July 2019 - June 2021
M.Sc. Engineering

IALab

Worked on DL models pertaining to Visual Reasoning, Adaptive Computation Time, Natural Language Processing and Action Recognition.

July 2017 - June 2021
Student Researcher

Pontificia Universidad Católica de Chile

Majored in Software Engineering (minor Data Science).
Graduated with distinction.

Jan. 2015 - Dec. 2019
B.Sc. Engineering

Publications

Spot The Ball: A Benchmark for Visual Social Inference
Neha Balamurugan and Sarah Wu and Adam Chun and Gabe Gaw and Cristobal Eyzaguirre and Tobias Gerstenberg
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2026
[arXiv] / [bibtex]

Taming generative video models for zero-shot optical flow extraction
Seungwoo Kim and Khai Loong Aw and Klemen Kotar and Cristobal Eyzaguirre and Wanhee Lee and Yunong Liu and Jared Watrous and Stefan Stojanov and Juan Carlos Niebles and Jiajun Wu and Daniel L. K. Yamins
Neural Information Processing Systems (NeurIPS), 2025
[arXiv] / [GitHub] / [Webpage] / [bibtex]

Understanding Complexity in VideoQA via Visual Program Generation
Cristobal Eyzaguirre Igor Vasiljevic and Achal Dave and Jiajun Wu and Rares Andrei Ambrus and Thomas Kollar and Juan Carlos Niebles and Pavel Tokmakov
International Conference on Machine Learning (ICML), 2025
[arXiv] / [GitHub] / [Webpage] / [bibtex]

T*: Re-thinking Temporal Search for Long-Form Video Understanding
Jinhui Ye* and Zihan Wang* and Haosen Sun and Keshigeyan Chandrasegaran and Zane Durante and Cristobal Eyzaguirre and Yonatan Bisk and Juan Carlos Niebles and Ehsan Adeli and Li Fei-Fei and Jiajun Wu and Manling Li
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2025
[arXiv] / [GitHub] / [Webpage] / [bibtex]

Streaming Detection of Queried Event Start
Cristobal Eyzaguirre and Eric Tang and Shyamal Buch and Adrien Gaidon and Jiajun Wu and Juan Carlos Niebles
Neural Information Processing Systems (NeurIPS), 2024
[arXiv] / [GitHub] / [Webpage] / [Blog] / [Poster] / [bibtex]

IKEA Manuals at Work: 4D Grounding of Assembly Instructions on Internet Videos
Yunong Liu and Cristobal Eyzaguirre and Manling Li and Shubh Khanna and Juan Carlos Niebles and Vineeth Ravi and Saumitra Mishra and Weiyu Liu and Jiajun Wu
Neural Information Processing Systems (NeurIPS), 2024
[arXiv] / [GitHub] / [Webpage] / [bibtex]

HourVideo: 1-Hour Video-Language Understanding
Keshigeyan Chandrasegaran and Agrim Gupta and Lea M. Hadzic and Taran Kota and Jimming He and Cristobal Eyzaguirre and Zane Durante and Manling Li and Jiajun Wu and Fei-Fei Li
Neural Information Processing Systems (NeurIPS), 2024
[arXiv] / [GitHub] / [Webpage] / [bibtex]

When Do Universal Image Jailbreaks Transfer Between Vision-Language Models?
Rylan Schaeffer and Dan Valentine and Luke Bailey and James Chua and Cristobal Eyzaguirre and Zane Durante and Joe Benton and Brando Miranda and Henry Sleight and John Hughes and Rajashree Agrawal and Mrinank Sharma and Scott Emmons and Sanmi Koyejo and Ethan Perez
ICLR 2025. The work also appeared at NeurIPS workshops RBFM (Best Paper), AdvMLFrontiers (Oral), Red Teaming GenAI (Oral), SoLaR (Spotlight), Safe & Trustworthy Agents., 2025
[arXiv] / [bibtex]

Self-Supervised Learning of Representations for Space Generates Multi-Modular Grid Cells
Rylan Schaeffer and Mikail Khona and Tzuhsuan Ma and Cristobal Eyzaguirre and Sanmi Koyejo and Ila Fiete
Conference on Neural Information Processing Systems (NeurIPS), 2023
[arXiv] / [Webpage] / [Poster] / [bibtex]

Revisiting the ''Video'' in Video-Language Understanding
Shyamal Buch and Cristobal Eyzaguirre and Adrien Gaidon and Jiajun Wu and Li Fei-Fei and Juan Carlos Niebles
(Oral) IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2022
[arXiv] / [GitHub] / [Webpage] / [Blog] / [Poster] / [bibtex]

DACT-BERT Differentiable Adaptive Computation Time for an Efficient BERT Inference
Cristobal Eyzaguirre and Felipe del Rio and Vladimir Araujo and Alvaro Soto
(ACL Workshop) NLP Power! The First Workshop on Efficient Benchmarking in NLP, 2022
[arXiv] / [GitHub] / [bibtex]

Differentiable Adaptive Computation Time for Visual Reasoning
Cristobal Eyzaguirre and A. Soto
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2020
[arXiv] / [GitHub] / [Blog] / [bibtex]


Minor Contributions in Other Fields

Lagrangian scale decomposition via the graph Fourier transform
MacMillan, Theodore and Ouellette, Nicholas T
The Eurographics Association, 2022
[pdf] / [bibtex]

Evaluating Interactive Comparison Techniques in a Multiclass Density Map for Visual Crime Analytics
Svicarovic, Lukas and Parra, Denis and Lobo, Maria Jesus
The Eurographics Association, 2020
[pdf] / [bibtex]