Gün Kaynar

Gün Kaynar

I am a Ph.D. student in the School of Computer Science at Carnegie Mellon University. I am fortunate to be advised by Carl Kingsford, the Herbert A. Simon Professor of Computer Science.

My research develops methods across machine learning, AI for science, and computational biology. In machine learning, I work on reasoning distillation for diffusion language models, study failure modes and reliability of tool-using LLM agents, and develop techniques for deployment-time shortcut mitigation and imbalanced classification. In collaboration with Arm, I develop simulation frameworks, orchestration and scheduling systems, error handling, and agentic supervision for programmable cloud laboratories, enabling AI agents to autonomously operate lab instruments. Separately, I'm part of Anthropic's Model Hardware Standard (MHS) research preview, letting AI agents safely operate lab equipment and run experiments autonomously. I also study reinforcement learning for multi-objective sequence optimization problems in biology, and build structure-aware models that integrate biophysical simulation with machine learning for modeling translation; in mRNA therapeutics, coding sequences optimized by these methods undergo experimental production and validation. Additionally, I work on clinical informatics, including adverse event prediction from electronic health records, differentially private synthetic EHR generation, and disease monitoring in transplant patients.

Previously, I earned my M.Sc. in Computer Science at Bilkent University, advised by A. Ercument Cicek, where I worked on biologically informed neural networks, NMR modeling, metabolomics, and CNV calling. I earned my B.Sc. at Bogazici University and spent a semester as an exchange student at Universitat de Barcelona.

Outside of research, I enjoy climbing, playing blues on guitar, watching festival movies, and reading.

News

seq2ribo published in Bioinformatics — July 2026
seq2ribo is now published in Bioinformatics.

seq2ribo accepted to ISMB 2026 — April 2026
Accepted as a proceedings paper; will be published in Bioinformatics.

Awarded the Randy Bryant Endowed Fellowship Fund — April 2026
School of Computer Science, Carnegie Mellon University.

Two questions accepted to the Humanity's Last Exam dataset — April 2026
Two of my questions on computational biology were accepted into the Humanity's Last Exam dataset.

seq2ribo accepted to RECOMB 2026 — March 2026
seq2ribo has been accepted to be presented at RECOMB 2026.

CMLH 2025 Fellowship Applicant Reviewer — December 2025
I served as a reviewer for fellowship applicants at the Center for Machine Learning and Health, Carnegie Mellon University.

Started the Ph.D. Certificate in Ethics and Artificial Intelligence — August 2025
School of Computer Science and the Dietrich College of Humanities and Social Sciences, Carnegie Mellon University.

Completed the HPC & Data Science Summer Institute — August 2025
Attended the summer institute at the San Diego Supercomputer Center, UC San Diego.

LYCEUM accepted to ISMB 2025 — April 2025
Accepted as a proceedings paper and published in Bioinformatics.

Started my Ph.D. at Carnegie Mellon University — August 2024
Joined the School of Computer Science.

ECOLE featured in Nature Communications Editors' Highlights — July 2024
Selected among the 50 best papers in the Biotechnology and Methods category of Nature Communications.

Graduated from Bilkent University — July 2024
Earned my M.Sc. in Computer Science.

Blog Posts

Anthropic Model Hardware Standard Preview: An AI Agent Ran an Automated Lab Experiment From Scratch — August 27, 2026
Notes on building with Anthropic's Model Hardware Standard research preview.

Publications

seq2ribo: Structure-aware integration of machine learning and simulation to predict ribosome location profiles from RNA sequences
Gün Kaynar, C Kingsford
Bioinformatics 42 (Supplement_1), 2026

Augmenting Electronic Health Records for Adverse Event Detection
Gün Kaynar, Z You, RD Boyce, T Yakoh, C Kingsford
medRxiv, 2026

Models Know Their Shortcuts: Deployment-Time Shortcut Mitigation
J Li, S Tang, Gün Kaynar, S Du, C Kingsford
AACL-IJCNLP, 2026

CodonRL: Multi-Objective Codon Sequence Optimization Using Demonstration-Guided Reinforcement Learning
S Du, Gün Kaynar, J Li, Z You, S Tang, C Kingsford
bioRxiv, 2026

LYCEUM: learning to call copy number variants on low-coverage ancient genomes
MA Yilmaz*, AA Ceylan*, Gün Kaynar*, AE Cicek
Bioinformatics 41 (Supplement_1), 2025
*Equal contribution

ECOLE: Learning to call copy number variants on whole exome sequencing data
B Mandiracioglu, F Ozden, Gün Kaynar, MA Yilmaz, C Alkan, AE Cicek
Nature Communications 15 (1), 2024

PiDeeL: metabolic pathway-informed deep learning model for survival analysis and pathological classification of gliomas
Gün Kaynar, D Cakmakci, C Bund, J Todeschi, IJ Namer, AE Cicek
Bioinformatics 39 (11), 2023

Targeted metabolomics analyses for brain tumor margin assessment during surgery
D Cakmakci*, Gün Kaynar*, C Bund, M Piotto, F Proust, IJ Namer, AE Cicek
Bioinformatics 38 (12), 2022
*Equal contribution

Teaching

02-750: Automation of Scientific Research
Teaching Assistant, Carnegie Mellon University, Spring 2026

02-613: Algorithms and Advanced Data Structures
Teaching Assistant, Carnegie Mellon University, Fall 2025