Research Projects
The Institute for Data Science & Artificial Intelligence brings together experts from across disciplines to foster interdisciplinary collaboration. Our core and affiliated members pursue both fundamental and applied research in artificial intelligence and big data technologies, developing data-driven solutions to challenges in fields as diverse as health, finance, engineering, and the basic sciences.
Below is a list of research projects currently underway at the Institute, reflecting the breadth of academic interests and scholarly contributions among our faculty and researchers. Some of these projects also offer scholarship or employment opportunities for students. We invite you to explore the list and reach out to the relevant faculty members for more information.
Project Name: On-Device LLM Assistant for Real-Time and Dynamic API Execution
Managed By: Asst. Prof. Mustafa Taha Koçyiğit
Start Date: 10/06/2025
Supported By: BAP - YÖK ADP
Project No: 24-50183
Project Name: LLM & RL-Based Disaster Management Support System Utilizing Multiple Information Sources
Turkish Title: LLM ve RL Tabanlı Çoklu Bilgi Kaynağı Kullanan Afet Yönetim Destek Sistemi
Managed By: Hüseyin Oktay Altun
Supported By: BAP - YÖK ADP
Start Date: 26/06/2025
Status: Ongoing
Project Summary: This project develops an LLM- and RL-based disaster management decision support system to improve crisis management during natural disasters. The goal is to provide a robust digital solution that helps minimize impact and improve the efficiency of rescue operations.
Project Name: Automatic Processing and Analysis of Kazasker Ruznamçe Records with Digital Methods
Turkish Title: “Kazasker Ruznamçe Kayıtlarının Dijital Yöntemlerle İşlenmesi ve Analizi”
Managed By: Metin Bilgin
Project Team: Şaziye Betül Özateş
Supported By: TÜBİTAK 3005
Start Date: 13/11/2024
End Date: 13/11/2026
Status: Ongoing
Students Supported Through Project Funding: 1 PhD Student (Scholarship Holder), 1 Master's Student (Scholarship Holder) — Dilara Zeynep Gürer, Aysenur Ercan
Project Summary: This project aims to extract information from the Anatolian Kazasker Ruznamçe registers (covering 1660–1895) using automatic and semi-automatic methods. The extracted data will then be linked and used to analyze the economic, administrative, and institutional structures of the Ottoman Empire during this period. For the first time, it will be possible to study an uninterrupted period of approximately 250 years.
Project No: 3231475
Project Name: Development of an Artificial Intelligence Model Supported by Molecular Biomarkers for Subtyping Breast Cancer
Turkish Title: Meme Kanserinin Alt Tiplendirmesine Yönelik Moleküler Biyobelirteçler İle Desteklenen Yapay Zeka Modelinin Geliştirilmesi
Managed By: Hüseyin Oktay Altun
Supported By: TÜBİTAK TEYDEB 1501
Start Date: 01/07/2024
End Date: 30/06/2026
Status: Ongoing
Project Summary: This project develops a multimodal and explainable decision support approach for breast cancer diagnosis and molecular subtyping by modeling three imaging modalities (mammography, ultrasonography, MRI) with AI and integrating histopathological data plus gene-expression analyses specific to molecular subtypes. In total, five categories of data are combined to support diagnosis and subtyping.
Project Name: Increasing Productivity and Quality in Agriculture with Reinforcement Learning Based PAW Production System
Managed By: Asst. Prof. Hüseyin Oktay Altun
Start Date: 01/06/2024
End Date: 31/05/2026
Supported By: TUBITAK TEYDEB 1501
Project No: 19985SUP
Project Name: A Multi-Agent Deep Reinforcement Learning Approach to Optimization of Friction Damper Systems for Seismically Excited Buildings
Turkish Title: “Sismik Uyarımlı Binalar için Sürtünme Sönümleyici Sistemlerinin Optimizasyonuna Yönelik Çok Etmenli Derin Takviye Öğrenme Yaklaşımı”
Managed By: Ercan Atam
Supported By: BAP
Start Date: 03/05/2024
End Date: 03/05/2026
Status: Ongoing
Students Supported Through Project Funding: 1 Master's Student (Employed Research Staff) — Brusk Serhat Solgun
Project Summary: The optimal placement of friction dampers in multi-story buildings against severe earthquakes is a large-scale, non-convex, multi-objective optimization problem. This project develops a multi-agent deep reinforcement learning approach to tackle this challenging problem.
Project Name: Robust Dense Correspondence via Multi-Modal Fusion and Transformers
Managed By: Asst. Prof. Mustafa Taha Koçyiğit
Start Date: 06/03/2025
End Date: 06/03/2026
Supported By: BAP
Project No: 20042SUP
Project Name: Operational and Strategic Analysis in Tactical Environment Simulations with Reinforcement Learning-Based Virtual Agents
Turkish Title: Pekiştirmeli Öğrenme Tabanlı Sanal Varlıklar ile Taktik Çevre Simülasyonlarında Harekat ve Strateji Analizi
Managed By: Hüseyin Oktay Altun
Supported By: BAP
Start Date: 17/12/2024
End Date: 17/12/2025
Status: Ongoing
Project Summary: This project aims to optimize RL-based behavior models so that virtual entities can develop more effective strategies against environmental factors. The project also investigates whether failures in a given scenario stem from scenario-design factors.
Project No: 20029SUP
Project Name: Deep Learning-based Exploration of Linguistic Structures and Semantic Entities in Historical Turkish Texts
Turkish Title: “Tarihi Türkçe Metinlerde Dilbilimsel Yapıların ve Anlamsal Varlıkların Derin Öğrenme Tabanlı Keşfi”
Managed By: Şaziye Betül Özateş
Supported By: BAP
Start Date: 11/10/2024
End Date: 11/10/2025
Status: Completed
Students Supported Through Project Funding: 1 Master's Student (Employed Research Staff) — Zeynep Yıldız
Project Summary: This project aims to perform structural and semantic analyses of Ottoman Turkish documents using deep learning techniques. To address the lack of labeled data, a treebank and a named entity recognition dataset will be manually created.
Project Name: Enhancing Tactical Environment Simulators with Reinforcement Learning by Enabling Virtual Entities to Adapt to Strategic Commands Using Large Language Models and Discovering New Permanent Skills to Enrich the Action Space
Managed By: Asst. Prof. Hüseyin Oktay Altun
Start Date: 01/08/2024
End Date: 01/08/2025
Supported By: TUBITAK BILGEM
Project No: 104582
Project Name: TÜBİTAK BİLGEM — AI-Supported Advanced Control Techniques for Variable-Mass UAVs
Turkish Title: “TÜBİTAK BİLGEM-Değişken Kütleli İHA’lar için Yapay Zeka Destekli İleri Kontrol Teknikleri”
Managed By: Ercan Atam
Supported By: TÜBİTAK BİLGEM
Start Date: 01/08/2024
End Date: 01/08/2025
Status: Completed
Students Supported Through Project Funding: 1 PhD student (Employed Research Staff)
Project Summary: This project develops two advanced control methods, supported by AI, for UAVs with changing mass. The methods will be tested on a high-fidelity simulation model and the results will be compared.
Project Name: Development of Artificial Intelligence-Based Cost Estimation Software for Automotive Suspension and Steering Parts
Managed By: Murat Arslanoğlu
Project Team: Assoc. Prof. Şener Özönder, Asst. Prof. Mustafa Taha Koçyiğit, Asst. Prof. Hüseyin Oktay Altun
Start Date: 1/1/2024
End Date: 30/06/2025
Supported By: TUBITAK TEYDEB - 1711
Students Supported Through Project Funding: 1 Master's Student (Scholarship Holder)
Project Name: Detection of Vulnerabilities in Post-Quantum Cryptography Algorithms with Deep Learning
Managed By: Assoc. Prof. Şener Özönder
Start Date: 03/06/2024
End Date: 30/05/2025
Supported By: TUBITAK BILGEM
Project Name: Multi-modal End-to-end All-Weather Autonomous Driving
Turkish Title: “Çoklu-modalite Uçtan-uca Tüm Hava Koşullarında Otonom Sürüş”
Managed By: Taha Koçyiğit
Supported By: TÜBİTAK BİLGEM
Start Date: 03/06/2024
End Date: 30/05/2025
Status: Completed
Project Summary: This work proposes an AI-based driving algorithm enhanced with a sensor-fusion network. The algorithm will use reliable sensor signals by evaluating noisy and limited data, enabling safer driving. It is designed to handle challenging traffic conditions such as merging lanes, lane changes, and right-of-way decisions at intersections.
Project No: 19766S
Project Name: High-Performance AI-based Algorithms in Drug Discovery
Turkish Title: İlaç Keşfinde Yapay Zeka Destekli Yüksek Performanslı Algoritmalar
Managed By: Şener Özönder
Supported By: Boğaziçi University, BAP
Start Date: 09/03/2023
End Date: 10/07/2024
Status: Completed
Project Summary: The aim of this project is to accelerate the computer-aided drug design workflow used to identify drug candidates that dock to target proteins.
Project No: 120F354
Project Name: New Nano-materials Discovery with AI and Computational Quantum Mechanics
Turkish Title: Yapay Zeka ve Hesaplamalı Kuantum Mekaniği ile Yeni Nano Malzemelerin Keşfi
Managed By: Şener Özönder
Supported By: TÜBİTAK 3501 KARIYER
Period: 2022 – 2024
Status: Completed
Students Supported Through Project Funding: 2 Master's Students (Scholarship Holders) — Mevlüt Köz, İpek Yenda
Project Summary: This project aims to accelerate computational chemistry workflows (e.g., density functional theory) using AI models.
Project No: EHPC-DEV-2025D06-088
Project Name: EuroHPC-Multi-Modal Large Language Model for Cross-Modal Reasoning and Generation through Scalable Dataset Annotation and Contrastive Alignment
Managed By: Taha Koçyiğit
Project Team: Suayb S. Arslan, Mehmet Turan
Supported By: European Union
Start Date: 20/06/2025
End Date: 20/06/2026
Status: Ongoing
Project Summary: This work first aims to generate large-scale synthetic multi-modal data and then adapt them into formats that can be used as inputs for large language models (LLMs). This enables LLM reasoning and analysis capabilities to be applied over data sources such as infrared, radar, LiDAR, and millimeter-wave modalities.
Project Name: Development of a Deep Learning-Based Data Expansion Tool for Predicting the Efficacy of Antibacterial Nanoparticles Used in Nanomedicine
Managed By: Emine Sümeyra Turali-Emre
Researcher: Şaziye Betül Özateş
Supported By: TÜBİTAK BİLGEM
Status: Completed
Project Summary: This work aims to build an AI-based database to understand the interaction between antibacterial nanoparticles (NP) and bacteria, accelerating the development of new antimicrobial agents. A comprehensive database will be developed to systematically analyze NP–bacteria relationships and to serve as a foundation for machine learning models predicting NP efficacy.
Project Name: AI-Assisted Drug Discovery Platform
Turkish Title: Yapay Zeka Destekli İlaç Keşfi Platformu Geliştirilmesi
Managed By: Şener Özönder
Supported By: TÜBİTAK TEYDEB 1501
Start Date: 01/04/2025
End Date: 31/03/2027
Status: Ongoing
Project Summary: This project aims to develop an algorithm and GPU-based software that can accelerate computational drug discovery pipelines by at least 1000×.
Project Name: On Device AI Agents: Revolutionizing Real-time API Interactions with Privacy Preserving Intelligence
Turkish Title: Cihaz Üzerinde Yapay Zeka Ajanları: Gizliliği Koruyan Zeka ile Gerçek Zamanlı API Etkileşimlerini Dönüştürmek
Managed By: Şener Özönder
Researcher (Co-lead): Taha Koçyiğit
Supported By: EU Project (FFPlus & EuroHPC)
Start Date: 01/01/2025
End Date: 31/10/2025
Status: Completed
Project Summary: This project develops secure, privacy-focused AI agents that can run on mobile devices and use tools via API/MCP.
Project Name: BatteLLIGENCE - AI and Cloud Based Battery Monitoring Platform
Turkish Title: BataryaZEKASI - Yapay Zeka ve Bulut Tabanlı Batarya İzleme Platformu
Managed By: Şener Özönder
Supported By: TÜBİTAK TEYDEB 1512
Start Date: 01/02/2023
End Date: 31/07/2024
Status: Completed
Project Summary: The BataryaZEKASI platform estimates battery health and remaining useful life, enabling electric vehicle manufacturers to predict maintenance and replacement times with sufficient accuracy, supporting more accurate total cost calculations and warranty planning.
Project No: 123K484
Project Name: The Dissemination of Knowledge in the Classical Islamic World: Rihlas
Managed By: Bekir Kuzudişli
Project Team: Taha Koçyiğit
Supported By: TÜBİTAK 1001
Start Date: 01/02/2025
End Date: 20/11/2026
Status: Ongoing
Project Summary: This project aims to analyze historical works and biographies about scholarly journeys (rihlas) conducted between Hijri years 150 and 300 to collect hadith transmissions, and to answer research questions related to these travels. Using AI, information will be extracted from dozens of historical sources to build a knowledge graph. Retrieval-augmented question answering (RAG) and knowledge-graph–augmented question answering (Graph-RAG) methods will be tested.
Project No: EHPC-AI-2024A04-095
Project Name: Revolutionizing Industrial Enzymes through Generative AI
Turkish Title: “Üretken Yapay Zekâ ile Endüstriyel Enzim Devrimi”
Managed By: İbrahim Çağrı Kurt
Project Team: Taha Koçyiğit
Supported By: EU Project
Start Date: 01/01/2025
End Date: 30/12/2025
Status: Ongoing
Project Summary: In this project, we train large-scale protein language models using metagenomic protein sequences. We then curate datasets of industrially relevant enzymes from metagenomic data and fine-tune our models on these datasets. We generate synthetic proteins and evaluate them first through computational simulations and then by producing and testing them in laboratory conditions. Our goal is to discover novel industrial enzymes that do not have direct natural counterparts, yet are more active, scalable, and robust compared to existing alternatives in nature.
Partnership: in partnership with Asst. Prof. Ibrahim Cagri Kurt
Program: Funded by EuroHPC (part of Horizon Europe), 2025
Project No: 20204SUP
Project Name: Robust Dense Correspondence via Multi-Modal Fusion and Transformers
Turkish Title: “Çok Modlu Birleştirme ve Transformer’lar ile Sağlam Yoğun Eşleştirme”
Managed By: Taha Koçyiğit
Supported By: BAP (Boğaziçi University)
Start Date: 06/03/2025
End Date: 06/03/2026
Status: Completed
Project Summary: Dense correspondence estimation between images is a fundamental component for computer vision tasks such as structure-from-motion, multi-view stereo and visual localization. Achieving robust correspondences under varying illumination, viewpoint changes and occlusions remains a major challenge. This project proposes a method that leverages multi-modal sensor fusion and transformer-based architectures to improve robustness and accuracy.
Project Name: Continual Pre-training of Large Language Models for Historical Language Understanding
Start Date: 2024
Status: Ongoing
Project Name: Simultaneous Reinforcement Learning-based Optimization of the Number, Distribution, and Adaptive Slip Forces of Friction Dampers for Earthquake-Resistant Buildings
Turkish Title: “Depreme Dayanıklı Binalar İçin Sürtünme Sönümleyici Sayısı, Dağılımı ve Adaptif Kayma Kuvvetinin Pekiştirmeli Öğrenme Tabanlı Eşzamanlı Optimizasyonu”
Managed By: Ercan Atam
Supported By: TÜBİTAK BİLGEM
Start Date: 01/09/2025
End Date: 30/05/2026
Status: Ongoing
Students Supported Through Project Funding: 1 Master's student, 1 PhD student
Project Summary: This project aims to simultaneously optimize the number of friction dampers in multi-story buildings, their inter-story distribution, and the adaptive slip force of each damper.
Project Name: Cross-API Agents for Reliable End-to-End Workflow Automation
Turkish Title: Güvenilir Uçtan Uca İş Akışı Otomasyonu için Çapraz-API Ajanları
Supported By: EuroHPC
Start Date: 2026
Status: Ongoing
Project Name: MProDockAI: Exploration of potential inhibitors of SARS-CoV-2 main protease via docking, molecular dynamics and SMILES transformer AI model
Turkish Title: MProDockAI: SARS-CoV-2 ana proteazının potansiyel inhibitörlerinin kenetleme, moleküler dinamik ve SMILES transformer yapay zeka modeli aracılığıyla araştırılması
Managed By: Şener Özönder
Supported By: EuroHPC
Start Date: 29/08/2025
End Date: 30/08/2026
Status: Ongoing
Project Name: Ring-Aware Transformer Attacks to Post-Quantum Cryptography Algorithms
Turkish Title: Kuantum Sonrası Kriptografi Algoritmalarına Halka-Farkında Dönüştürücü Saldırıları
Managed By: Şener Özönder
Supported By: TÜBİTAK BİLGEM
Start Date: 2025
Status: Ongoing
Students Supported Through Project Funding: 1 Master's Student — Hümeyra Yüksel
Project Name: EUPilot - Development of Open Source Software and Hardware AI Accelerator Chips for HPC Centers
Turkish Title: EUPilot - Hesaplama Merkezleri için Açık Kaynaklı Yapay Zeka Hızlandırıcı Yazılımı ve Donanımı Geliştirme
Supported By: EU Project
Start Date: 2024
Project Name: Detection of Vulnerabilities in Post-Quantum Cryptography Algorithms with Deep Learning
Turkish Title: Derin Öğrenme ile Kuantum Sonrası Kriptografi Algoritmalarının Kırılganlıklarının Tespiti
Managed By: Şener Özönder
Supported By: TÜBİTAK BİLGEM
Start Date: 03/06/2024
End Date: 30/05/2025
Status: Completed
Project Summary: This project tests transformer-based approaches for extracting secret keys to identify vulnerabilities of lattice-based post-quantum cryptography algorithms against deep learning. By evaluating a wide range of transformer models, the project aims to identify safe regions in the parameter space.
Project No: 3231475
Project Name: Development of an Artificial Intelligence Model Supported by Molecular Biomarkers for Subtyping Breast Cancer
Turkish Title: Meme Kanserinin Alt Tiplendirmesine Yönelik Moleküler Biyobelirteçler İle Desteklenen Yapay Zeka Modelinin Geliştirilmesi
Managed By: Hüseyin Oktay Altun
Supported By: TÜBİTAK TEYDEB 1501
Start Date: 01/07/2024
End Date: 30/06/2026
Status: Ongoing
Project Summary: This project develops a multimodal and explainable decision support approach for breast cancer diagnosis and molecular subtyping by modeling three imaging modalities (mammography, ultrasonography, MRI) with AI and integrating histopathological data plus gene-expression analyses specific to molecular subtypes. In total, five categories of data are combined to support diagnosis and subtyping.
Project No: 3251247
Project Name: Designing Virtual Entity Behavior Models that Can Reason with Large Language Models and Reinforcement Learning and Can Be Shaped by Domain Experts
Turkish Title: Büyük Dil Modelleri Ve Pekiştirmeli Öğrenme İle Akıl Yürütebilen Ve Alan Uzmanları Tarafından Şekillendirilebilen Sanal Varlık Davranış Modeli Tasarımı
Managed By: Hüseyin Oktay Altun
Supported By: TÜBİTAK
Start Date: 01/10/2025
End Date: 30/09/2027
Status: Ongoing
Students Supported Through Project Funding: 1 PhD student, 4 high school (or younger) students
Project Summary: This project aims to develop domain-knowledgeable AI models for military simulations by integrating LLMs with RL to enable reasoning, robust decision-making, and expert-guided shaping of virtual-entity behavior.
