I am very interested in AI research. I share some thoughts on papers on LinkedIn from time to time. I completed my PhD in Machine Learning, focusing on the limitations of current ML models.
My research outcome
- I invented a test-time defense prompt technique (model reprogramming), being the first to my knowledge. Link
- I found that multimodal latent diffusions are different from small perturbations. Link
- I improved adversarial examples outlier detection by creating a dataset through open-source contribution (OpenOOD) and wrote light-weight detectors with extended evaluations. Link
Research interests: Model robustness, reasoning, deepfake, test-time learning, autonomous systems, foundation models, pattern recognition, large-scale AI, multimodality, self-improving architectures, and AGI.
📫 How to reach me: My mail
- 🔭 I’m currently working on transformers and foundation models.
- 🌱 I’m currently learning agengtic AI and evolutionary AI.
- 👯 I’m looking to collaborate on ML research projects. (Collaborate with other AI-obsessed people and create something awsome.)
- 🤔 I’m looking for mock interview practice. Link
- ⚡ Fun fact: I won a tournament in kickboxing.