Bruno Fonkeng is a Computer Science Ph.D. student at the University of North Dakota, advised by Dr. Jielun Zhang. His work sits at the intersection of probabilistic machine learning, cybersecurity, and high-performance computing, with a focus on generative models and adaptive defense for modern cyber threats. He is particularly interested in using GANs, VAEs, diffusion models, and related architectures to simulate realistic APT behavior on networks, malware traffic, and cloud infrastructures, and then coupling these simulations with reinforcement-learning–based defenders that can learn to respond in real time. Through teaching and research assistantships, he contributes to UND’s initiatives in advanced computing, applied AI, and secure systems engineering.

Before beginning his doctoral studies, Bruno earned a B.Sc. in Physics from the University of Bamenda, Cameroon, where he developed strong analytical and quantitative skills that sparked his interest in scientific computing and AI. The rigorous mathematical training from physics now underpins his research in probabilistic modeling, statistical reasoning, and algorithmic optimization.

At UND, he has completed graduate coursework spanning machine learning, applied machine learning, artificial intelligence, data visualization, predictive modeling, application layer security, ethical hacking, security for cloud computing, high-performance computing & paradigms, and computer forensics, along with key mathematics courses in linear algebra and statistical theory. He is actively deepening his expertise in HPC, probabilistic modeling, and formal methods to support his dissertation work.

Bruno’s current research explores probabilistic generative models for cybersecurity, including synthetic cyber-traffic and malware image generation, data augmentation for intrusion detection, and robust anomaly detection. His long-term goal is to develop a probabilistic simulation-to-defense framework that can generate realistic, large-scale cyber data and train self-improving defensive agents for cloud environments. By combining rigorous statistical modeling, scalable experimentation on HPC systems, and practical systems thinking, he aims to help close the gap between academic ML models and deployable, trustworthy security tools that protect real organizations.

What I Do Best

Probabilistic Machine Learning 

I design and train probabilistic and generative models—GANs, VAEs, diffusion models, and Bayesian networks—for real-world problems in cybersecurity and reliability. My work emphasizes rigorous evaluation, uncertainty estimation, and reproducible experiments, with a focus on models that can scale on HPC systems and actually work in practice.

Data Engineering & ML Pipelines

I build the data and infrastructure that surround models: ETL/ELT pipelines, streaming and API integrations, experiment scaffolds, and deployment workflows. From malware and cyber-traffic datasets to ROI pipelines for industry partners, I focus on making ML systems robust, observable, and production-ready.

August 2025 – Present

August 2023 – May 2025

October 2017 – August 2020