BIO

I am currently pursuing an MS in Data Science at Columbia University. I received my bachelor’s in Neuroscience from BU, during which I conducted an honors thesis on the dynamics of temporal representation in episodic memory under the supervision of Dr. Joe McGuire. In Joe’s lab, I worked with fMRI data from the 13 TB Natural Scenes Dataset to analyze retrieval activity patterns in the medial temporal lobe. In my free time, I enjoy biking, cooking, and visiting museums.

Work in the lab

During the summer of 2024, I worked with Grant to adapt the multimodal ProSmith model for olfaction-related tasks. I tested and refined this model using olfaction datasets such as M2OR and DoOR. ProSmith was evaluated on both classification and regression tasks and was used to generate embeddings for small molecules and proteins utilizing the underlying transformers ESM1b and ChemBERT.

Publications and presentations

Publications

• McConachie, G.D., Duniec, E., Guerina, F.V., Younger, M.A., DePasquale, B. (2026). Low rank adaptation of chemical foundation models generates effective odorant representations. International Conference on Learning Representations (ICLR) 2026. preprint

Presentations

• McConachie, G.D., Duniec, E., Guerina, F.V., Younger, M.A., DePasquale, B. (2025). Learning the Language of Smell: Foundation Models for Protein-Odor Interactions. Frontiers in NeuroAI, Kempner Institute at Harvard University, Cambridge, MA.
• McConachie, G.D., Duniec, E., Guerina, F.V., Younger, M.A., DePasquale, B. (2024). Improved Odor-Receptor Interaction Predictions via Self-Supervised Learning. Neuroscience to Artificially Intelligent Systems (NAISys), Cold Spring Harbor Laboratory, NY. poster