About
Islam Tayeb
I'm a senior studying Computer Science at Duke University in Durham, NC. I want AI to be useful for chemistry, so I build the data and eval infrastructure that scientific agents and foundation models need. I'm also interested in systems and verifiers. I currently work with Philip Romero and Kevin Yang to extract data at scale and study how it affects chemistry foundation models.
Before that, I used to do organic chemistry research on porous polymers and published a few papers. Then, I transitioned to software engineering to experience startups and worked at Soff (YC S24), now Westgate Supply.
I was born and raised in Egypt, but moved to Taif, Saudi Arabia during high school. I like to play Tetris and Monkeytype in my free time.
In high school, I worked as a graphic designer for an esports team. I also played osu! competitively and designed a few skins (500K+ downloads).
- Presented at the AI Scientist Summer Workshop at Microsoft Research in BostonAug 2026
- Accepted into the Anthropic AI for Science programJun 2026
Experience
Research
- Duke University Philip Romero + Kevin Yang (Microsoft Research)Aug 2025 - Present
Extracting data at scale and studying how it affects chemistry foundation models
Grants: Anthropic AI for Science Program
Outputs: AI Scientist Summer Workshop (Poster)
- Duke University Matthew LentzMay 2026 - Jul 2026
Test-passing code by construction, building on Trellis
- Duke University Navid NaderiAlizadehDec 2024 - Apr 2025
Continual learning for antibody affinity prediction
Outputs: Preprint
- KFUPM Mahmoud AbdelnabyJul 2022 - Aug 2024
Polymer property prediction and synthesis for CO2 capture
Collabs & Grants: Saudi Aramco
Outputs: 3 publications @ Q1 journals in applied ML, organic synthesis, and a review
Engineering
- Soff (YC S24; now Westgate Supply)May 2025 - Oct 2025
Agentic sales intelligence for manufacturers, employee #2
- Duke Impact Investing GroupSep 2024 - May 2025
Non-linear RNA-seq analysis for CRISPR experiments
Collabs: Life Edit Therapeutics
- Duke Institute for Health InnovationJun 2024 - Aug 2024
Automated literature review agents + VTE detection
Teaching
- Operating Systems Matthew LentzAug 2026 - Dec 2026
Leading a discussion section + office hours
- Computer Systems Matthew LentzJan 2026 - May 2026
Led a discussion section + office hours
- Organic Chemistry I SAGE TutorJan 2025 - May 2025
Led a study group, saw kids quit pre-med as the semester went
Open Source
- RaycastOct 2025 - Present
Made Helium extension, browser tab control (2.2K+ downloads)
Contributions: Messages (54K) + Paste as Plain Text (28K)
Selected Publications
- Machine learning for predicting and optimizing the CO2 uptake in porous organic polymersHamid Zentou, Ali Tayeb, Islam Tayeb, Mahmoud AbdelnabyResearch Article / Journal of Environmental Chemical EngineeringSep 2025
- Primal dual continual learning for robust antibody designIslam Tayeb, Navid NaderiAlizadehPre-printMay 2025
- Post-synthetic modification of UiO-66 analogue metal-organic framework as potential solid sorbent for direct air captureMahmoud Abdelnaby, Islam Tayeb, Ahmed Alloush, Hussain Alyosef, Aljazi Alnoaimi, Mostafa Zeama, Mohammed Mohammed, Sagheer OnaiziResearch Article / Journal of CO2 UtilizationJan 2024
Writing
- Becoming the Service (or 10→1 vs. 1→0)opinionOct 20260.8K words (3 mins)
- Agent Memory FidelitytechnicalJun 20262.0K words (8 mins)
- osu! and FingerspitzengefühlpersonalMar 20260.7K words (3 mins)
