Machine Learning Researcher | Drug Discovery | Cheminformatics
PhD candidate at the University of Vienna, specializing in machine learning for drug discovery and out-of-distribution generalization. Interdisciplinary background spanning Biomedical Engineering (MSc, Sharif University) and Mechanical Engineering (BSc, Amirkabir University).
- Out-of-distribution generalization and robustness for molecular property and activity prediction
- Test-time training and domain adaptation in chemical space
- Graph neural networks for chemistry and drug discovery
- Generative models for drug design (Diffusion, Flow-based models)
- High-performance scientific tooling in Rust (PDB/SDF parsers, molecular tokenizers)
Research
| Project | Description |
|---|---|
| ALineMol | Benchmarking ML models on out-of-distribution molecular data |
| THEMAP | Task hardness estimation for molecular activity prediction |
| molax | Molecular active learning with JAX |
Educational
| Project | Description |
|---|---|
| GNNs-For-Chemists | GNN implementations from scratch for chemists |
| Transformers-For-Chemists | Transformers from scratch for chemists (sister course to GNNs-For-Chemists) |
| bayesoptimol | Interactive Jupyter Book on Bayesian optimization and active learning in drug discovery |
Rust tooling & data
| Project | Description |
|---|---|
| pdbrust | Rust library for parsing PDB files |
| sdfrust | Fast, pure-Rust parser for SDF and MOL2 chemical structure files |
| rustmolbpe | BPE tokenizer for molecular SMILES in Rust with Python bindings |
| chembl-pdb-linker | Curated protein-ligand dataset linking ChEMBL bioactivity with PDB co-crystal structures |
| lincs_processing | Filtering and reshaping LINCS L1000 data by cell line, compound, dose, and time |