Projects
Selected research and engineering projects in NLP and machine learning.
LLM Collusion and Identifiability in Code Generation
Investigates whether large language models can recognize or attribute code authored by themselves or other LLMs.
Bias Toward Languages Studied in NLP Peer Reviews
First systematic study of language-of-study bias in NLP peer review, finding that non-English papers face bias rates roughly 40× higher than English-only ones.
Multimodal Framework for Detecting Implicit Hate Speech on Social Media
Designed a hybrid framework combining text encoders (BERT, LLaMA) with vision backbones (Swin Transformer, ViT) to detect subtle hate speech.
Named Entity Recognition in Turkish Tweets (Yildirim Platform)
Developed an annotation platform (Yildirim) for labeling named entities in Turkish tweets.
Procedural Text Understanding via Python Code Representation
Explored Python code representation as an intermediate reasoning layer for LLMs.
Intent Prediction and Slot Filling with Alexa MASSIVE Dataset
Trained multilingual intent and slot-filling models on the Alexa MASSIVE dataset using PyTorch and custom hyper-parameters.
Shilling Attack Detection in Recommender Systems
Investigated detection of profile injection (shilling) attacks in recommender systems.