Dinko Franceschi
PhD Student, Harvard University
Associated Researcher, Broad Institute of MIT and Harvard
200 Longwood Avenue
Boston, MA 02115
About
Dinko Franceschi’s research centers on mechanistic interpretability and the development of safe AI models for biology. At Harvard Medical School, he leads a project using sparse autoencoders to discover interpretable features in DNA sequence and mechanics.
Previously, Dinko worked as an ML Engineer at the Broad Institute. He was awarded a machine learning research Fellowship as a Visiting Scholar the Max Planck Institute for Software Systems in Saarbrücken, Germany. Prior to his PhD in Systems Biology at Harvard, Dinko studied mathematics at Columbia University.
Recent Publications
Proteome-wide model for human disease genetics R. Orenbuch, A. Kollasch, H. Spinner, C. Shearer, T. Hopf, L. Van Niekerk, D. Franceschi, M. Dias, J. Frazer, D. Marks
ProteinGym: Large-Scale Benchmarks for Protein Design and Fitness Prediction.
P. Notin, A. Kollasch, D. Ritter, L. Van Niekerk, S. Paul, H. Spinner, N. Rollins, A. Shaw, R. Weitzman, J. Frazer, M. Dias, D. Franceschi, et al.
Research Interests
- Human Genetics
- Protein Design
- Non-coding DNA