EquiMon

My NSF GRFP research proposal represents the phase of carrying forward the insights from my DivShift project into a broader, long-term framework. Drawing directly on my experiences designing machine learning approaches for biodiversity monitoring, this proposal envisions a new research direction, EquiMon, that integrates human well-being metrics and de-biasing strategies to make ecological modeling more equitable. Having never written a formal research proposal before, I concentrated on articulating the societal and scientific rationale of the project while demonstrating how it builds on the biases, domain insights, and sustainability lessons learned from my previous work.

In drafting this proposal, I translated ML methods (like contrastive learning and distributionally robust optimization) into language that shows clear real-world payoff, particularly for decision-makers in funding and policy contexts. Communicating the human impact (e.g., equity in low-resource regions) alongside ML frameworks proved central, ensuring that each objective felt relevant, urgent, and feasible to technical reviewers. Much like my previous sustainability projects, the proposal focuses on the importance of threading domain-specific knowledge (ecology and social well-being) into an ML-centric narrative. This not only elevated the proposal's clarity but also highlighted the broader social implications of focusing AI research on underrepresented communities and ecosystems.