Research Engineer, Computer Vision @ MBZUAI · Abu Dhabi, UAE
Website · Google Scholar · LinkedIn · X
I’m a Research Engineer at MBZUAI (Computer Vision).
My work focuses on vision-language learning and temporal / spatiotemporal understanding, with recent projects spanning VLM adaptation and 3D scene understanding.
I enjoy building research codebases that are readable, reproducible, and easy to extend.
- Vision-Language Models (VLMs), large-scale multimodal learning
- Temporal understanding, spatiotemporal representation learning
- Self-supervised learning
- 3D perception for autonomous systems
- microCLIP: Unsupervised CLIP adaptation for fine-grained image classification
Repo: https://github.com/sathiiii/microCLIP - DPA: Dual Prototypes Alignment for unsupervised adaptation of VLMs
Repo: https://github.com/sathiiii/DPA - S2TPVFormer: Spatiotemporal tri-perspective view for 3D semantic occupancy prediction
Repo: https://github.com/cepdnaclk/e17-4yp-S2TPVFormer
- Projects: https://sathiiii.github.io/projects/
- CV: https://sathiiii.github.io/assets/docs/Sathira_Silva_CV_latest.pdf
- Email: sathira.silva@mbzuai.ac.ae
- Personal: sathirasofte@gmail.com
