United States
Building tomorrow’s AI.
Developed AI workflows for InVEST ecosystem models(NATCAP), integrating LLM based agents with environmental data systems to automate existing models to enhance web-based modeling, and support sustainable development research.
Designed and implemented the Static Node project for Amazon Bedrock Flows, enabling low-latency, cost-efficient, and consistent performance across generative AI workflows with modular reusability throughout the Flows service; additionally, supported the development of Amazon Bedrock AgentCore to streamline the deployment and management of generative AI agents.
Led the development of the official Global Methane Abatement Solutions Tracker (G-MAST) and supported evidence-based policy, research, and stakeholder engagement. Collaborated on the development of an automated pipeline for streamlining Systematic Literature Reviews, enhancing research workflow efficiency.
Contributed on adding Parameter Efficient Kronecker Product based adapters to the K, Q, V, O matrices of the attention module in personalised Text-to-Image Generative Models(SDXL 1.0 and SD 2.1) that are ∼35% more efficient than SOTA, while generating images with higher fidelity and text-alignment. Accepted @ WACV 2025 Supervisor: Dr. Pin-Yu Chen