Dogukan Aksu, Ph.D.

I am an AI Engineering Lead at AMS, where I lead the architecture and delivery of enterprise agentic AI systems. My work focuses on turning rapidly evolving AI capabilities into reliable, scalable, and practical solutions for real-world business needs.

I work across the AI engineering lifecycle, from LLM applications, RAG and GraphRAG to agent architecture, tool use, MCP, orchestration, evaluation, observability, security, governance, and deployment. I enjoy working at the intersection of technology and business, understanding what clients and stakeholders actually need, translating those needs into sound AI architectures, and working with teams to turn ideas into systems that can be deployed and operated in the real world.

A particular area of focus for me is what it takes to make AI systems genuinely reliable in production. Building an AI system is only the beginning. I am interested in how we evaluate whether agents are behaving as intended, understand and diagnose failures, observe system behaviour in production, and use meaningful feedback to continuously improve performance. I care about building AI systems that are not only capable, but also measurable, observable, secure, maintainable, and trustworthy.

Before joining AMS, I was a Postdoctoral Research Fellow at the Centre for Secure Information Technologies (CSIT), Queen’s University Belfast (QUB), where I worked closely with Assoc. Prof. Ihsen Alouani and Prof. Jesus Martinez del Rincon. My research focused on Federated Learning, AI security, and privacy-preserving machine learning, particularly gradient leakage and gradient inversion attacks. I worked on privacy-preserving Federated Learning pipelines and evaluated attacks including DLG, iDLG, and GRNN.

At QUB, I also contributed to TruBrain, an international collaboration involving EPFL and Sorbonne University, with Intel and Northrop Grumman as industrial partners. The project focused on secure neuromorphic computing and Spiking Neural Networks (SNNs). I also represented the consortium at CHIST-ERA.

Previously, I was a Principal Applied Scientist at the Scientific and Technological Research Council of Turkiye (TUBITAK), Informatics and Information Security Research Center (BILGEM), where I worked on Generative AI and advanced machine learning projects, including deepfake detection, LLM-based voice assistants, and generative AI for identity verification.

Earlier in my career, I was an AI Tech Lead and Senior AI Research Engineer in the AI Enablement Department at Huawei, leading Click-Through Rate (CTR) prediction projects for large-scale advertising systems. During this time, I was honored with the Sparkling Star and Consumer Cloud Service President awards. I also led a TUBITAK University-Industry collaboration project between Huawei and Middle East Technical University (METU), where I worked with Prof. Ismail Hakki Toroslu, Prof. Pinar Karagoz, and Prof. Ismail Sengor Altingovde.

I hold a B.Sc., M.Sc., and Ph.D. in Computer Engineering from Istanbul University. During my M.Sc. and Ph.D. studies, I focused on machine learning approaches to intrusion detection systems under the supervision of Prof. M. Ali Aydin, while also serving as a Research and Teaching Assistant. During my undergraduate studies, I also founded an AI start-up company, Aksu Informatics, funded by a techno-initiative capital program from the Ministry of Science, Industry, and Technology. I developed AI software solutions to support people with disabilities, including people with ALS.

My background combines AI research, large-scale machine learning, cybersecurity, and AI engineering leadership. Today, I focus on turning advanced AI capabilities into robust production systems and developing the engineering practices needed to make them reliable, measurable, secure, and continuously improvable.