I’m a final-year Ph.D. candidate (expected April 2027) in the Durrant Lab at the University of Pittsburgh, with five years of drug discovery research at the intersection of computational chemistry and machine learning. My work centers on building large, quality-graded protein–ligand datasets for ML training and testing ML models against experimental data to find where they succeed and where they take shortcuts. Before that, I trained in computational biology and genomics, and I still bring that toolkit to collaborative projects. I’m now seeking industry roles in computational drug discovery, ML for chemistry, and cheminformatics / scientific software, and I’m open to relocation. Here is how I got here.
Collected and analyzed EEG and facial-expression data for the Lumiere concentration-detection project, then interpreted variants and ran enrichment analysis (g:Profiler, KEGG) for an Alzheimer’s GWAS study.
Research
Undergraduate Thesis — AML drug sensitivity
Giza, Egypt
Trained Random Forest models on patient RNA-seq profiles to predict AML sensitivity to five ERBB inhibitors, identifying markers linked to differential drug response. Advised by Dr. Eman Badr; graduated Cum Laude and on the Provost’s Honors Roll.
Transition
Moved to the U.S. to pursue drug discovery
Pittsburgh, PA
My thesis clarified the gap between finding genomic markers and knowing how a patient actually responds to a drug. To work on that gap, I moved to Pittsburgh for a PhD in Molecular, Cellular & Developmental Biology, focused on computational structural biology and drug design.
Research
First-year rotations
Pittsburgh, PA
Completed three ten-week rotations spanning the structural-biology spectrum: the VanDemark Lab (wet-lab structural biology — testing small-molecule inhibitors of the profilin–actin interaction by fluorescence anisotropy), the Durrant Lab (computational structural biology and drug design), and the Levin Lab (comparative genomics).
Transition
Joined the Durrant Lab
Pittsburgh, PA
Chose the Durrant Lab for my dissertation, focusing on structure-based drug design and open-source scientific software. Began developing LIGNOVA, an automated pipeline that docks bioactive PubChem compounds against experimental protein structures to build large-scale, quality-graded datasets for machine learning — now over 258 million docked poses across 5.86 million protein–ligand pairs (243,805 ligands, 11,268 structures).
Research
Passed the comprehensive examination
Pittsburgh, PA
Defended an independent research proposal in an NIH F31 format to my committee, clearing the PhD comprehensive exam.
Research
Presenting LIGNOVA at Gordon Research Conferences
West Dover, VT & Portland, ME
Presented early versions of LIGNOVA as posters at the Computer-Aided Drug Design GRC (2023) and the Computational Chemistry GRC (2024). See publications.
Research
Genomics collaboration — PNAS 2025
Pittsburgh, PA
Extended my Levin Lab rotation into a co-first-author study of how a hypermutable hotspot drives rapid recognition-gene evolution in Dictyostelium. Led the computational genomics, genome assembly, QC, and targeted annotation across ten chromosome-length genomes. Published in PNAS (2025).
Led recitations and weekly office hours for ~200 students and mentored 15 undergraduate TAs. Nominated by students for the Elizabeth Baranger Excellence in Teaching Award in 2024 and 2026.
Teaching
Undergraduate Research Mentor — TECBio REU
Pittsburgh, PA
Co-mentored an undergraduate on a ten-week project applying AlphaFold2 to explore HIV protease conformations for drug design, culminating in a research poster.
Research
Scaling LIGNOVA and stress-testing ML scoring
Pittsburgh, PA & Cambridge, MA
Earned a competitive NSF ACCESS allocation (Purdue Anvil + TAMU ACES) to scale LIGNOVA, and used the dataset to stress-test a deep-learning affinity model, exposing a molecular-size shortcut and blindness to activity cliffs. Presented the full pipeline at the Computational Medicinal Chemistry School in October, supported by a GPSG travel grant.
Present
Advanced to PhD Candidacy
Pittsburgh, PA
Proposed my dissertation work to my committee and was admitted to candidacy. Now wrapping up my PhD (expected April 2027) and seeking industry roles in computational drug discovery, ML for chemistry, and cheminformatics / scientific software.
LIGNOVA: An automated, open-source pipeline that has generated 258M+ quality-graded docked poses across 5.86M protein–ligand pairs, 89% of them labeled with experimental affinities, for ML training.
ML evaluation: Benchmarked a deep-learning affinity model against 5M+ experimental labels across 10,937 proteins and exposed a molecular-size shortcut and activity-cliff blindness, confirmed with permutation and bootstrap controls.
Docking benchmark: Tested whether consensus rescoring (ComBind) can replace explicit-water modeling in Glide pose prediction on 215 X-ray complexes. Manuscript in preparation.
Reverse virtual screen: Led a structure-based search (Foldseek, AlphaFold3, docking) for yeast arginine sensors that sequence homology misses; a subset of my 4 candidates was experimentally validated by the O’Donnell Lab.
Comparative genomics: Led the computational genomics for a co-first-author PNAS paper on how Dictyostelium recognition genes evolve so rapidly.