Rohil Rao

About

I hold an M.Sc. in Computer Science from the Universität Bonn, where I specialized in Artificial Intelligence. My research focuses on systematic evaluation of deep neural networks, vision‑language models, and reliable AI for autonomous systems.

Currently, I work with the AI Safety & Certification group at Fraunhofer IAIS, developing methods to detect systematic weaknesses in perception models and leveraging foundation models (e.g. CLIP, SAM, GPT‑4V) for data generation, augmentation, and annotation.

Publications

• Detecting Systematic Weaknesses in Vision Models along Predefined Human‑Understandable Dimensions
  S. S. Gannamaneni, R. P. Rao, M. Mock, M. Akila, S. Wrobel
  TMLR, 2025  · pdf
• Investigating CLIP for Metadata Generation in Autonomous Driving Datasets
  S. S. Gannamaneni, R. P. Rao et al.
  CVPR Workshops, 2023  · paper

Projects

3D MRI Segmentation UDA Entropy‑minimization domain adaptation pipeline that aligns 3D brain‑MRI segmentation models across scanners & hospitals.

Self‑Supervised Video Prediction ConvLSTM model trained without labels to predict future frames and synthesize short video clips with temporal consistency.

Explainable Fake News Detection BERT‑based classifier with attention heat‑maps and Streamlit dashboard for transparent fake‑news detection.

CV

Download my full résumé here (PDF).

Contact

Email: rohil93@gmail.com
GitHub: @rohilrao
LinkedIn: rohilrao