Theo Sternlieb

I am a machine learning scientist at Dyno Therapeutics where I work on structure-based machine learning for capsid discovery. I'm broadly interested in supervising protein structure foundation models with alternative data modalities and applying them to protein design. I obtained my MA in Computer Science at Wesleyan University in the Thayer Lab where I worked on self supervised learning for graph generative models of small molecules as well as RL finetuning for target specific small molecule optimization. Prior to that I completed my BA in Mathematics at Wesleyan University.

LinkedIn | Resume | GitHub

Papers

• Low-n OpenFold fine-tuning improves peptide design without additional structures
  Sternlieb, Theodore, Jakub Otwinowski, and Jeffrey Chan. "Low-n OpenFold fine-tuning improves peptide design without additional structures" (2024)
De Novo Short Linear Motif (SLiM) Discovery With AlphaFold-Multimer
Sternlieb, Theodore, Abhishaike Mahajan, Davian Ho, and Jeffrey Chan. "De Novo Short Linear Motif (SLiM) Discovery With AlphaFold-Multimer." (2023).
• Beyond the training set: an intuitive method for detecting distribution shift in model-based optimization
  Damani, Farhan, David H. Brookes, Theodore Sternlieb, et al. "Beyond the training set: an intuitive method for detecting distribution shift in model-based optimization." (2023).
• Target Specific Drug Design with Deep RL
  Sternlieb, Theodore. "Target Specific Drug Design with Deep RL." (2022).