AI Protein Structure Function Predictor

Orbio AI

Hybrid · $160k – $200k / year · Full-time · Posted 4 months ago

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

About the role

AlphaFold told us what a protein looks like. What it doesn't tell us is what the protein does. Structure-to-function prediction — given a folded protein structure, predict its enzymatic activity, binding partners, subcellular localization, and disease associations — is the next major challenge in computational biology, and the one most directly relevant to drug discovery. We partner with 12 pharmaceutical companies on target identification and lead optimization. Our structure-to-function models are used to prioritize which protein variants to synthesize and test in the lab, reducing the experimental search space from thousands of variants to dozens. That means our false negatives cost months and millions of dollars. You'll lead structure-to-function prediction modeling: designing architectures that extract function-relevant features from 3D protein structures, integrating multi-modal signals from sequence, evolutionary conservation, and structural context, and building the uncertainty quantification that tells drug hunters how confident to be in each prediction. The role is hybrid — 2 days per week in our Cambridge MA office for wet lab collaboration with the biology team.

Responsibilities

Design structure-to-function prediction models integrating 3D structural and evolutionary sequence features
Build multi-modal fusion architectures combining sequence embeddings with structural graph representations
Implement uncertainty quantification providing calibrated confidence on function predictions
Validate models against experimental assay results with wet lab biology collaborators
Develop active learning pipeline prioritizing protein variants for experimental synthesis
Publish methodology in structural bioinformatics or computational biology venues

Requirements

7+ years in computational biology, bioinformatics, or ML for drug discovery
Deep expertise in protein structure representation learning and geometric deep learning
Strong Python and PyTorch skills with experience using AlphaFold or ESM-family embeddings
Understanding of protein biochemistry sufficient to evaluate prediction validity
Track record of ML models that translate to experimental outcomes