Scholar
Ankit Sonthalia
Google Scholar ID: XUB_XqIAAAAJ
Tübingen AI Center, Universität Tübingen
Citations & Impact
Citations
10
H-index
2
i10-index
0
Publications
4
Co-authors
8
list available
Publications
Resume (English only)
Academic Achievements
- - Publication: 'On the rankability of visual embeddings', with Arnas Uselis, Seong Joon Oh, 2025 arXiv pre-print
- - Publication: 'Do Deep Neural Network Solutions Form a Star Domain?', with Alexander Rubinstein, Ehsan Abbasnejad, Seong Joon Oh, 2025 International Conference on Learning Representations (ICLR)
Research Experience
- - PhD Student: Scalable Trustworthy AI, March 2024 – Present, Tübingen, Germany
- - Research Software Engineer: Tübingen AI Center, August 2022 – February 2024, Tübingen, Germany
- - Research Assistant: Deutsches Forschungszentrum für Künstliche Intelligenz, February 2022 – June 2022, Kaiserslautern, Germany
Education
- - M.Sc. Mathematics, 2022, Birla Institute of Technology & Science (BITS), Pilani
- - B.E. Computer Science, 2022, Birla Institute of Technology & Science (BITS), Pilani
Background
- Research Interests: Trustworthy Machine Learning, Vision-language models, Embedding spaces, Parameter spaces. Professional Field: Scalable Trustworthy AI. Biography: Currently a second-year PhD student in Seong Joon Oh's group. Previously involved in various research activities during undergraduate studies, internships, and as a research software engineer. Recently interested in understanding how and why neural networks work, with a long-term goal of injecting principles of causality into AI.
Miscellany
- Feel free to reach out over email or LinkedIn to share ideas or seek help.
