Resource Constrained Machine Learning, Multi-Armed Bandits

ECE 236A - Linear Programming

ECE 236A - Linear Programming

Using and developing AI models to predict possible diseases

Worked under the supervision of Prof. Sinem Cöleri on Visible Light Communication. Built a simulation suite for comparison of selected Visible Light Positioning SOTA papers, and applied Cramer-Rao Lower Bound on the methods to compare theoretical estimation bounds.

- Worked under the supervision of Asst. Prof. Fatma Güney on developing and implementing a novel object detection model that utilizes from spatio-temporal information through Faster R-CNN backbone and Transformer model. - Used MOT17Det dataset in the project and inspected other detection/tracking datasets for further use.- Implemented the model from scratch using PyTorch, DGL, and GluonCV, and performed more than 100 experiments on Amazon EC2 instances.- Participated in and contributed to weekly paper discussions.

Worked under the supervision of Prof. Sinem Çöleri on a research project for autonomous vehicles to understand headings and positions of other vehicles on the road using Visible Light Communication and Angle of Arrival (AoA).

Image Processing and AI Computers Unit

Communications and Information Technologies - Image Processing and Digital Design Research & Development Team

CSBridge aims to give rising high school juniors, especially girls, coding experience. It is a program based on the curriculum of Stanford’s introductory programming methodology class, CS106A. As a section leader, I conducted problem sessions, helped students during lab hours and gave feedback to their daily programming assignments.

COMP130 - KOLT is an office that aims to help students understand the course material better via different programs. In peer tutor program, a student, who got A or A+ from a course, gives reviews and helps students about lecture material and exam preparation in weekly sessions.