Sarah Ibrahimi
Researcher
Computer Vision & Multimodal Learning
I obtained my PhD degree in Computer Vision and Multimodal Learning at the University of Amsterdam in 2024, under the supervision of Prof. Dr. Marcel Worring and Dr. Nanne van Noord. My background is in Computer Science & Engineering (MSc), Mathematics (BSc), and Theater and Dance Studies (BA).
My main areas of interests are Multimodal Learning, Visual Retrieval tasks, AI & Creativity and more recently Hyperbolic Learning.
I am now on the job market and open to research positions starting from Summer or Fall 2025.
News
- 12//2024: I defended my PhD thesis, called Learning from Real-World Data Challenges for Similarity Search at the University of Amsterdam! You can find my thesis here.
- 09/2024: I started as a Student Researcher in the Learning2Perf team at Google Cloud AI in Sunnyvale, California. During this internship, I will work on using multimodal LLMs to improve ML productivity and efficiency.
- 06/2024: Our work Intriguing Properties of Hyperbolic Embeddings in Vision-Language Models has been accepted at TMLR. This work is the last paper of my PhD.
- 08/2023: I started as an Applied Scientist Intern in the New Initiatives team of Amazon Just Walk Out, where I'm working with Greg Hager and Austin Reiter. During this internship, I will work on uncertainty quantification on vision-language models.
- 07/2023: Our work Audio-Enhanced Text-to-Video Retrieval using Text-Conditioned Feature Alignment has been accepted as an oral at ICCV 2023. This work is the result of my internship at Amazon Prime Video.
- 08/2022: I started as an Applied Scientist Intern in the Multimodal Video Understanding team of Amazon Prime Video, led by Mohamed Omar. During this internship, I will work on audio-enhanced text-video retrieval.
- 10/2021: Our work Learning with Label Noise for Image Retrieval by Selecting Interactions has been accepted at WACV 2022. This work is the result of my internship at Naver Labs Europe.
- 10/2021: Our work Inside Out Visual Place Recognition has been accepted at BMVC 2021. In this work we present the new task Inside Out Visual Place Recognition where the goal is to localize indoor images together with a new dataset.
- 10/2021: Our work Instance-level Recognition for Artworks: The Met Dataset has been accepted at NeurIPS 2021, Datasets and Benchmark Track.
- 09/2021: My work Composition, Performance and Evaluation: A Dance Education Framework for AI Systems has been accepted at the International Conference on Computational Creativity 2021.
- 07/2021: I finished my internship at Naver Labs Europe which resulted in two papers that are currently under submission.
- 06/2021: My work 'A Comparison between Dance Generation Algorithms and Choreographers' has been accepted at the CVPR Workshop on Ethical Considerations in Creative applications of Computer Vision 2021.
- 01/2021: I started as a Research Intern at Naver Labs Europe in the Machine Learning and Optimization team where I'll be working on optimizing interaction losses and learning with noisy labels.
- 05/2020: Our work on Detecting CNN-Generated Facial Images in Real-World Scenarios has been accepted at the CVPR Workshop on Media Forensics 2020.
- 05/2020: Our work Conditional Image Generation and Manipulation for User-Specified Content has been accepted at the CVPR AI for Content Creation Workshop 2020.
- 10/2019: I presented our work on Deep Metric Learning for Cross-Domain Fashion Instance Retrieval at the ICCV Workshop on Computer Vision for Fashion, Art and Design.
- 07/2019: Our work on Multimodal Semantic Matching has been accepted at ACM Multimedia 2019 as a demo paper.
