Foundational AI models hold immense potential to revolutionize diverse fields yet challenges in customization and adaptability persist. This presentation highlights innovative strategies to harness these models for impactful solutions.

I begin with MolGen-Transformer, a molecular foundation model leveraging transfer learning and supervised fine-tuning to accelerate discovery, as presented at NeurIPS 2024 AI4MAT. At Corteva, I applied post-training customization using RLHF, integrating preference-based insights to tailor large language models for domain-specific needs.

During my Google X internship, I developed multi-agent collaboration frameworks using a Mix of Experts (MoE) architecture, enabling agents to dynamically solve complex tasks with multimodal data streams. Lastly, I introduce BioTrove, a NeurIPS 2024 spotlight paper featuring a research-grade, multimodal dataset advancing biodiversity research through rigorous curation and annotation.

This presentation demonstrates a systems-level approach to foundational AI, bridging innovative methodologies with cross-disciplinary applications in discovery, customization, and ecological research.

To add to calendar:
Click on:  https://wordpress.cels.anl.gov/cels-seminars/
Enter your credentials.
Search for your seminar 
Click “Add to calendar”