Kemal Özkırşehirli
MIT B.Sc. candidate · Computer Science and Engineering · Chemical Physics · Mathematics
Cambridge, MA, USA - kemalozk@mit.edu
Highlights:
- MIT B.Sc. candidate, expected May 2027; GPA 5.0/5.0; majors in Computer Science and Engineering, Chemical Physics, and Mathematics; minors in Philosophy and Writing.
- Current work spans protein foundation model red-teaming, ChromoGen-style 3D chromatin diffusion, autoimmune target discovery, TBXT CADD, ChemAgent-QSM, EVEdesign protein design, antibody RL, and ML-augmented renormalization group theory.
- Selected from 800+ applicants for PRISM 2026; selected as one of 22 worldwide Kupcinet-Getz Scholars; current and recent work across MIT, A*STAR, Columbia, Harvard Medical School, Weizmann, and Pedal AI.
- Medalist in the International Chemistry Olympiad and International Mendeleev Chemistry Olympiad; TÜBİTAK National Chemistry Olympiad silver medalist; World Science Scholar; Waterloo distinction recipient.
I build AI/ML, generative/diffusion-model, molecular simulation, free-energy, and inference systems that turn scientific complexity into computable representations for decision-making in drug discovery, protein design, chromatin biology, theoretical chemistry, and statistical mechanics.
Professional
About me - Professional - Education - Awards - Contact - CV (pdf)
Research - Protein Safety - CADD - Chromatin - Target Discovery - Quantum/Stat Mech
Teaching - 6.1200[J] - Berker Group - Mentoring
Writing - Essays held back for launch - AI for Science notes held back
Skills - Computational Stack - Languages - Humanities
AI/ML for Science - Selected Work
Protein Models and Safety
PRISM - adversarial evaluation, reliability checks, uncertainty calibration, and failure taxonomies for protein foundation models [code pending release]
EVEdesign - 21-person, 18-institution, 8-country open protein-design collaboration; uncertainty-aware multi-objective candidate selection [code pending release]
Antibody RL - structure-informed cross-attention transformer and PPO pipeline using OAS, SAbDab, IEDB, AF-Multimer/IGFold metrics
CADD and Drug Discovery
TBXT - 11-person chordoma/TBXT sprint; 2,274 compounds + 737 analogs + 30,000 BRICS recombinations to a 570-compound novelty-filtered pool
V2M - autoimmune genetics-to-mechanism target-discovery engine with single-cell foundation models, QTL/colocalization, and targetability checks [code pending release]
Pedal AI - LangGraph/FastAPI chemistry agents, retrosynthesis route trees, <50ms latency, 15,000+ daily API requests, 100,000+ predictions [code pending release]
Genome, Chromatin, Diffusion
ChromoGen V2 - conditional diffusion for small-molecule-perturbed single-cell 3D chromatin ensembles with Plate-C/Easy Dip-C/CHARM-style data [code pending release]
Biochemical Diffusion - PINN/neural-ODE stochastic simulation of nonequilibrium reaction-diffusion networks; Euler-Maruyama, Runge-Kutta, Gillespie SSA [code pending release]
Sequence-to-State - variant-effect scoring, Perturb-seq validation, counterfactual immune-state transport, and structure-aware target prioritization [code pending release]
Theory, Simulation, Agents
ChemAgent-QSM - LLM-guided quantum/statistical-mechanics pipelines using PySCF; TCFs, local order, relaxation, mobility, spectra
Renormalization Group - GNN/PyTorch Kadanoff RG framework for spin systems, fixed points, and phase boundaries [code pending release]
DFT → kMC - multi-scale computational organic chemistry pipeline; 20+ transition states, 500+ trajectories, 22% predicted yield optimization [code pending release]
Education and Awards
MIT - CS&E, Chemical Physics, Mathematics; Philosophy and Writing; GPA 5.0/5.0; expected May 2027
Olympiads - IChO, Mendeleev, TÜBİTAK, Istanbul Science Olympiads, CHEM13, Galois, Sir Isaac Newton distinctions
World Science Scholar - selected among 52 scholars from 22 countries for exceptional mathematical talent
Coursework and Toolchain
CS/AI - deep learning, NLP, algorithm engineering/design, sublinear algorithms, theory of computation, stochastic processes, computer systems
Chem/Physics/Math - advanced organic chemistry, probability, linear algebra, chemical physics, statistical mechanics, free-energy methods
Stack - Python, C/C++, CUDA, HPC, SLURM, PyTorch, RDKit, OpenMM/OpenFE, GNINA/Vina, Boltz-2, PySCF, GNNs, transformers
Teaching, Writing, Humanities
6.1200[J] - proof-intensive discrete math teaching fellow/grader for 250+ student MIT EECS cohort [code pending release]
Berker Teaching - mentored 200+ students while teaching condensed augmented chemistry and classical/quantum mechanics courses [code pending release]
Humanitas Core - theory of models, rhetoric, advanced essay/poetry, philosophy of language/religion/politics, metaphysics, memoir
My Pages
CV - full web CV with education, research, teaching, honors, skills, languages, interests, and hobbies
CV PDF - downloadable source CV
GitHub - public code profile and released repositories
LinkedIn - external profile
Email - contact
