Hi there, I'm Arian Amani👋
💬 I build AI that doesn't just predict biological states but learns to understand the underlying mechanisms of cellular response.
At the "Wellcome Sanger Institute", my research focuses on causal generative models for single-cell biology. The goal is true mechanistic clarity: using disentangled representation learning to separate a drug’s extrinsic effect from a cell’s intrinsic state. My work on "CellDISECT" introduced counterfactual reasoning to the field, allowing us to simulate "what if" scenarios for biological counterfactuals with unprecedented accuracy.
At "AI VIVO", I bridge the gap between high-level research and clinical utility. I design production-grade ML pipelines that integrate chemical structure with multi-omic assay data, turning theoretical breakthroughs into deployable drug discovery tools.
My background in "Computer Vision" left me with a permanent obsession with "Out-of-Distribution (OOD) generalization". In biology, where the "unseen" is the norm, I believe OOD robustness is the only way to build foundation models that actually generalize across patients and tissues.
- 🌱 I'm currently studying causality in deep learning, specifically, Causal Representation Learning.
- 🚀 My recent focus has been on creating cutting-edge Generative Models and pioneering Multi-Modal Contrastive Learning approaches.
- 🔬 Single-Cell Genomics, Drug Discovery, Drug Optimization, and Gene-Drug Interaction Analysis
- 📙 Check out my resume.
- ✒️ I might start to write articles on my blog regularly, for now there's a roadmap to starting Deep Learning
