Resume

Ege Cakar

Cambridge, MA

Education

Harvard University

AB Double Concentration in Statistics and Physics, GPA: 3.91/4.0
May 2027 • Cambridge, MA

Relevant Coursework:

COMPSCI 181: Machine Learning

COMPSCI 184: Reinforcement Learning

NEURO 240: Biological and Artificial Intelligence

STAT 110: Probability

STAT 111: Statistical Inference

COMPSCI 120: Algorithms and Their Limitations

PHYSICS 181: Statistical Mechanics and Thermodynamics

MATH 21A: Linear Algebra

MATH 21B: Multivariable Calculus

PHYSICS 143a: Quantum Mechanics

• John Harvard Scholar (top 10% of class)

American Collegiate Institute

Double diploma in the International Baccalaureate and the Turkish Curriculum
June 2023 • Izmir, Turkey

• Founded a physics education website; national award for mobile application

Technical Skills & Projects

Machine Learning Frameworks: PyTorch, numpy, TensorFlow, Gymnasium, PettingZoo, Wandb, TensorBoard
Programming: Python, C++, Mathematica, MATLAB, SLURM, GitHub
Design: Adobe Photoshop, Adobe Illustrator, Matplotlib, Seaborn

Reinforcement Learning Final Project

Eco Civilization MDP: A Multi-Agent Environment for Studying Sustainable Growth

• Developed a custom PettingZoo environment and trained agents using Multi Agent Proximal Policy Optimization (3-person team, 3000+ lines of code)
• Modeled collaboration and studied competitive growth vs. environmental preservation
• Currently expanding research and running further experiments

Biological and Artificial Intelligence Final Project

FocusCaption: A Novel Image Captioning Approach

• Used a saliency map generator to locate the point of interest in an image and separate it out
• Employed parallel CNNs to extract features from both full and focused images
• Observed improvements in counting image subjects and detailing finer features

Relevant Experience

The Kempner Institute for the Study of Natural and Artificial Intelligence

Student Researcher, Pehlevan Lab
June 2024 – Present • Cambridge, MA

• Investigated compositional Boolean calculation tasks across MLPs, LSTMs, Transformers
• Achieved consistent performance across widths with µp normalization in MLPs (KRANIUM fellow over Summer 2024)
• Explored self-supervised learning (SimCLR) for assessing representation quality (KURE fellow over Fall 2024)
• Investigating methods for transforming in-context knowledge in LLMs into persistent memory
• Ran large-scale experiments on the Kempner High Performance Cluster

Harvard University

Course Assistant – Theoretical Physics, Quantum Mechanics 1
September 2024 – Present • Cambridge, MA

• Led office hours and assisted students with advanced physics concepts
• Evaluated assignments and provided detailed feedback
• Collaborated with the instructor to identify and support struggling students