AI/ML Engineer
Trustworthy Intelligent Systems
Building systems that see, reason, act, and stay robust in production.
Ph.D. in Computer Science — University of Southern Mississippi, Hattiesburg, MS
I'm an AI/ML engineer and doctoral researcher building intelligent systems that work in the lab and hold up in production. I design and ship ML pipelines, multimodal applications, agentic workflows, and the cloud-native infrastructure—AWS, GCP, and Azure—that keeps them observable, scalable, and cost-efficient.
My Ph.D. at the University of Southern Mississippi centers on trustworthy intelligent systems: multimodal perception, agentic decision-making, and adversarial robustness—with an emphasis on systems that can be evaluated, defended, and deployed, not just trained.
I build in public and in production. Tracevox is an open-source LLM observability stack I contributed to—request tracing, cost analytics, and incident workflows for AI workloads at scale. From susceptibility mapping with Karst Intelligence Agent to serverless systems on AWS, I ship where model quality and operational reliability are measured—not assumed.
Ph.D. research focus
One umbrella, three pillars—how I approach building AI that teams can trust.
Multimodal & vision–language models
Systems that reason across vision, language, and other modalities—built for real inputs, not just benchmark leaderboards.
Agentic AI & reinforcement learning
Autonomous agents that plan, act, and learn from feedback—sequential decision-making wired for production constraints.
Adversarial robustness & safe deployment
Defending intelligent systems under attack, evaluating failure modes, and shipping guardrails teams can measure and trust.
Education
- University of Southern Mississippi
Ph.D. in Computer Science · Hattiesburg, MS - University of Colorado Boulder, CO — M.S. Computer Science
- Georgia State University, Atlanta, GA — M.S. Geosciences
- University of Ibadan, Ibadan, Nigeria — B.S. Geology
What I optimize for
Intelligent systems that resist failure, earn trust under adversarial pressure, and remain observable at scale. I aim for measurable, deployable outcomes—where multimodal models, reinforcement learning, and agentic architectures meet the engineering discipline to ship them safely.