Bruno Fonkeng

Conduct doctoral research on probabilistic generative modeling, synthetic cybersecurity data, malware classification, reliable evaluation, and reproducible high-performance research infrastructure.

• Develop GenCyberSynth, a framework for training, sampling, evaluating, comparing, and auditing synthetic cybersecurity-data generators.
• Evaluate GAN, VAE, diffusion, autoregressive, and probabilistic approaches using KID, MS-SSIM, balanced accuracy, macro F1, macro AUPRC, and downstream utility.
• Build reproducible HPC workflows with fixed seeds, experiment manifests, SLURM execution, artifact tracking, aggregation, figures, and publication-ready reports.
• Investigate selective augmentation policies, class-level performance, controlled generation budgets, and conditions under which synthetic data improves malware classification.

Support undergraduate instruction in systems programming, computer architecture, cybersecurity, computer networks, and information assurance.

• Assist students with C programming, pointers, memory, file I/O, processes, debugging, systems interfaces, and command-line development.
• Support coursework in computer architecture, data communications, computer and network security, and cybersecurity risk analysis.
• Review technical submissions and provide structured feedback on correctness, design, evidence, clarity, and secure engineering practices.

Contributed to an applied decision-support initiative involving ROI-calculation workflows, requirements analysis, validation, system design, and interdisciplinary technical documentation.

• Translated operational requirements into measurable inputs, calculations, outputs, and user-facing workflows.
• Supported applied research, validation, software-design decisions, and documentation for practical decision making.

Contributed to AI-training and evaluation tasks involving precise specifications, deterministic outputs, Python automation, Docker, structured JSON, and automated verification.

• Developed and refined technical tasks involving log processing, file-system operations, structured results, and test-driven verification.
• Evaluated instruction precision, reproducibility, expected output schemas, failure conditions, and agent behavior.