Research Scientist, Next-Generation Structural Biology & Atomistic Modeling

Valence Labs•Montreal, QC

11d•Hybrid

About The Position

We are seeking a Research Scientist with a hybrid research-engineering mindset to join our team. In this role, you will be at the forefront of developing generative architectures and foundation models that ground machine learning in real-world physical and biological discovery. You will focus on accelerating and improving the accuracy of molecular design and structural biology workflows—specifically targeting the intersection of physics-informed frameworks and data-driven ML to solve complex protein-ligand interaction challenges.

Requirements

PhD (or equivalent) with significant academic or industry research experience in machine learning applied to structural biology, atomistic modeling, or physical simulation.
Scientific knowledge of physics and chemistry, with a deep understanding of physical constraints and invariances in molecular systems.
Impactful research track record, including experience with equivariant models, generative modeling of molecular systems, or replacing traditional physics workflows (like ABFE) with ML-driven alternatives.
Strong technical and engineering skills, including proficiency in Python and the ability to build scalable, reproducible experiment pipelines.
Interdisciplinary empathy, with a proven ability to work effectively with medicinal chemists and biophysicists to ensure models solve real-world drug discovery problems.
Leadership and communication skills, including the ability to explain complex ideas clearly to both technical and non-technical stakeholders.

Responsibilities

Research and develop state-of-the-art architectures (e.g., flow matching, diffusion models, geometric deep learning) tailored to modeling protein-ligand interactions.
Develop hybrid approaches that integrate co-folding, molecular dynamics (MD), and experimental potency data to achieve high-resolution accuracy on novel targets.
Build and maintain ML systems capable of processing massive datasets, such as protein-ligand simulations, on high-performance compute clusters (BioHive).
Ensure ML predictions are biologically trustworthy and actionable by collaborating closely with drug discovery teams to reduce cycle periods and dead ends in lead optimization.
Publish findings in top-tier venues (e.g., NeurIPS, ICML, Nature, JACS) and contribute to the broader scientific community.