Researchers

Research areas: 1. Neuromophic Vision: event cameras, face and facial landmark detection, large-scale dataset creation 2. Remote Sensing: City Identification via Satellite data, City Sustainability prediciton via Satellite Data, Construction Sites Detection and Segmentation via Satellite data (Instance Segmentation)

Causal representation learning and foundation models for neuroscience

I research multimodal deep learning for mental-health assessment — detecting suicide risk, depression, and emotion from text, audio, and video using transformer ensembles and fine-tuned LLMs on real clinical and speech-emotion datasets.

My current research focus is centered on Applied Machine Learning, Deep Learning, and Multimodal AI, with a strong interest in document understanding, OCR, vision-language models, and LLM-based systems. I am currently exploring synthetic data generation for low-resource Turkic languages, especially for OCR and document recognition tasks. This includes building realistic multilingual document datasets, designing generation pipelines, and preparing data for fine-tuning modern OCR/VLM models. Another important direction of my work is the development of practical AI systems based on RAG, LLM agents, and multimodal reasoning. I am interested in turning research ideas into usable AI tools for real-world domains such as document processing, legal technology, education, and urban surveillance.

Research focus: any sort of evaluation methodologies for GenAI of any modality.

ACR on tote jazu, use of RL for better eval