Farhan Sadeek

I'm an undergraduate at Columbia University studying Computer Science with minors in Statistics and Applied Physics in the Fu Foundation School of Engineering and Applied Science. I spent a year at Dartmouth College, where I was advised by Peter Chin and Rahul Sarpeshkar, and the last two years of high school at The Ohio State University as a dual-enrolled student.

I'm generally interested in efficient inference, post-training, reinforcement learning, and diffusion models, with some work in mechanistic interpretability.

Research

	Learnable Path-Complex Features and Unified Path–Cellular Message Passing for Link Prediction Farhan Sadeek, Sie Hendrata Dharmawan, and Peter Chin Published, May 22, 2026 project page / paper We introduce position-aware learnable initialization for path-complex cells and a unified path–cellular complex for link prediction, combining path- and ring-based message passing.
	Simultaneous Illumination Normalization and Sparse Image Compression via Unfolded Recovery Farhan Sadeek Published, May 24, 2026 project page / paper The joint inverse-problem pipeline combines illumination normalization and sparse image compression, outperforming sequential recovery by up to 4.7 dB on textured scenes.