Senior Data Scientist - Protein Structure ML models

Career Category

Research

Job Description

Senior Data Scientist - Protein Structure ML models

Role Summary

The Senior Data Scientist - Protein Structure ML models will play a critical role in advancing AI-enabled protein and antibody design across Large Molecule Discovery (LMD). This role will focus on building, adapting, and validating machine learning models that predict protein function from structure, with particular emphasis on antibodies and antibody-like molecules.

Working at the intersection of machine learning, structural biology, protein engineering, and experimental discovery, this individual will also develop workflows that combine structure prediction, structure generation, and inverse-folding models into practical protein design pipelines. The role will partner closely with wet-lab scientists to guide assay design, generate high-value property data, and translate internal and externally available datasets into rigorous validation strategies.

This role is ideal for someone who enjoys developing technically rigorous geometric ML methods while remaining deeply connected to experimental validation and real-world biologics discovery needs.

Key Responsibilities

Protein Structure to Function Modeling

Build machine learning models, either developed in-house from scratch or adapted from existing examples, to predict protein function from structure.

Focus model development on antibodies and antibody-like molecules, including formats that require structure-aware modeling approaches for downstream property prediction.

Use internal and externally available structural, sequence, binding, and property datasets to evaluate and improve model performance.

Antibody Binding Model Validation

Design, own, and maintain validation tasks for assessing antibody binding models across internal and externally available datasets.

Establish well-curated benchmarks that support quality control, model comparison, and responsible onboarding of external / open-source ML models.

Translate validation results into actionable guidance for model selection, model improvement, and downstream protein design decisions.

Experimental Data Generation Partnership

Partner with wet-lab scientists and structural biologists to guide assays for collecting property data that can improve model training, validation, and decision-making.

Help define data collection strategies that connect experimental readouts to ML model objectives and biologics design hypotheses.

Work cross-functionally to interpret experimental outcomes and incorporate learnings into iterative model development workflows.

Integrated Protein Design Workflows

Develop workflows that chain or efficiently combine structure prediction, structure generation, and inverse-folding models for protein design.

Apply structure-aware ML tools to support de novo design and optimization of antibodies and related formats.

Contribute reusable workflows, validation practices, and technical standards that improve scalability and reproducibility of protein design pipelines.

Basic Qualifications

Bachelor’s degree in Computational Biology, Bioinformatics, Life Sciences, Computational Chemistry, Chemical Engineering, Materials Science, Data Science, or a related quantitative field and relevant professional experience.

Experience Requirements

Bachelor’s degree and 6+ years of relevant experience, OR

Master’s degree and 4+ years of relevant experience, OR

PhD

<>·

Preferred Qualifications

Programming and ML Frameworks

Fluent in Python and able to develop reliable, reusable scientific software for model development, data analysis, and workflow automation.

Hands-on experience with at least one modern deep learning framework, such as PyTorch or JAX.

Ability to work across research codebases, adapt existing implementations, and mature promising examples in the literature into robust internal workflows.

Protein Structure Modeling

Strong foundation and hands-on experience with protein structure modeling, particularly as applied to antibodies and antibody-like molecules.

Experience translating structural biology concepts into computational features, modeling assumptions, and validation strategies.

Structural modeling experience for small molecules is also welcome, especially where those skills are transferable to understanding protein surfaces.

Deep Learning for Protein Structures

Experience applying deep learning to protein structures, including graph neural networks, equivariance-aware methods, or related architectures.

Ability to evaluate structure-conditioned models for protein function, binding, designability, or related protein engineering objectives.

Comfortable working with noisy, sparse, or heterogeneous biological datasets and designing validation strategies that reflect practical discovery constraints.

Generative and Inverse Design Tools

Experienced in applying out-of-the-box generative protein structure tools, such as RFdiffusion or Boltzgen, to protein design questions.

Preferred experience extending or adding features to open source structure prediction or generative modeling tools.

Preferred experience using inverse design tools such as ProteinMPNN or related sequence-design methods within end-to-end protein design pipelines.

Scientific Collaboration and Communication

Ability to collaborate effectively with wet-lab scientists, protein engineers, structural biologists, ML researchers, and data engineering partners.

Strong communication skills for explaining model assumptions, validation results, and design recommendations to both technical and scientific audiences.

Strong problem-solving skills, scientific judgment, and attention to detail in the development and deployment of protein design workflows.

Success Measures

Success in this role will be demonstrated through:

Delivery of validated ML models that predict protein function from structure and support antibody and antibody-like molecule design.

Creation of robust validation tasks that provide clear quality control for internal and external antibody binding models.

Improved ability to guide wet-lab data generation toward high-value property measurements for model training and validation.

Effective integration of structure prediction, structure generation, and inverse-folding tools into reproducible protein design workflows.

Increased confidence in model-driven design decisions through well-documented benchmarks and validation practices.

Positive adoption of protein structure ML workflows by AI, computational biology, and experimental discovery teams.

Typical Candidate Profile

The ideal candidate combines strong machine learning expertise with practical experience in protein structure modeling and biologics discovery. They are comfortable building models, adapting frontier methods, and designing validation tasks that determine whether those models are useful for real discovery decisions.

Candidates may come from computational biology, machine learning, structural biology, protein engineering, bioinformatics, or related quantitative backgrounds. They are motivated by the opportunity to connect de novo protein design, antibody engineering, experimental validation, and scalable ML workflows into practical discovery capabilities.

Organizational Impact

This role contributes directly to de novo protein design efforts by developing and validating ML-enabled workflows that connect protein structure to function. By using internal datasets to guide de novo ML models, the Senior Data Scientist - Protein Structure ML models will help enable the design of exotic antibody-like formats and expand the range of biologics concepts that can be explored computationally. The role will also provide critical quality control for onboarding and pipeline use of external ML models through well-curated validation tasks, improving confidence in model-driven design decisions across Large Molecule Discovery.

.