I build AI agents for advertising and work on sensing and human-state modeling. These projects connect my research to questions about automation, information sharing, and human safety.
01 / AgentMark.ai · Chief Scientist
AI agents for advertising
At AgentMark.ai, I build AI agents for ads. My work connects conversational interfaces with campaign data, analysis, advertising management, content creation, and automation.
Advertising work connects strategy, data, content, and repeated operational tasks. Useful automation needs to bring those pieces together.
My contribution
I design and build the AI agent, connecting LLM capabilities with campaign data and advertising tools. This includes the conversational interface and its integration into advertising workflows.
The focus
Turning language and data into useful actions: analyzing information, managing advertising workflows, generating content, and supporting automation.
Research applications
02 / Research application
Resilient sensing & navigation
Applying deep and federated learning to GNSS interference detection and WiFi indoor localization. This work explores learning across heterogeneous data while reducing the need to pool raw observations.
The trustworthy-AI connection
Receivers and users see different conditions. Personalization helps address those differences, while distributed training changes what information needs to leave a device. I also investigated model quantization to reduce communication costs.
Connecting system features, physiological responses, and cybersickness through Bayesian networks. Probabilistic verification checks specified risk properties within the encoded model and evidence.
The trustworthy-AI connection
Prediction accuracy alone does not answer whether a specified risk is acceptable under a given condition. I connect human-state modeling with explicit safety queries, alongside research on engagement estimation under cognitive attack.
My GitHub is another place to explore my code and projects. Individual papers provide the methods, experiments, and publication details behind the research.