Principal Scientist, Molecular Perturbation Modeling

GSK•San Francisco, PA

10d•Hybrid

About The Position

As a (Senior) Principal Scientist in the Protein Design and Informatics (PDI) team, you will focus on translating biological mechanisms of disease to molecular mechanisms of therapeutics by integrating perturbation data to design new molecules that modulate disease phenotypes. The advancement in de novo design technologies has opened the doors to generate molecules to test biological hypotheses at scale, build up foundational data to predict new perturbation effects, accelerate the validation of disease intervention points, and drive therapeutic discovery campaigns. Included in the larger Data, Automation, and Predictive Sciences (DAPS) department, you will be the predictive engine for R&D, focusing on researching and embedding new methods to enable the vision of automation of the entire Design-Make-Test-Analyze cycle, driving Lab-in-an-Automated-Loop frameworks from target discovery to the clinic - all stages of a therapeutic project. You’ll have the opportunity to work in close partnership with many departments across GSK, developing and fostering a high-performing team culture of collaboration, curiosity, consistency, agility, quality, peer review, and continuous improvement with a relentless focus on creating medicines for patients. We create a place where people can grow, be their best, be safe, and feel welcome, valued and included. We offer a competitive salary, an annual bonus based on company performance, healthcare and wellbeing programmes, pension plan membership, and shares and savings programme. We embrace modern work practises; our Performance with Choice programme offers a hybrid working model, empowering you to find the optimal balance between remote and in-office work. Discover more about our company wide benefits and life at GSK on our webpage Life at GSK | GSK In this role you may Work to generate, validate, and integrate multimodal generative AIML models for the de novo design and multi-objective optimization of tool and therapeutic molecules, such as miniproteins, antibodies, antigens, peptides, ADCs, and oligonucleotides. Guide molecular perturbation experiments that validate mechanisms of disease and show reversal of disease phenotypes and signatures. Build and exploit agent-orchestrated, integrated Design-Make-Test-Analyze cycles with automated experimental platforms, generating quality data at scale needed for project-specific and foundational models. Identify and advocate for the opportunities afforded by scientific computation and platform automation and driving therapeutic project plans with predictive technologies. Collaborate with external groups to further develop protein engineering computational methods. Predict and evaluate potential disease intervention points for their probability of success to be therapeutically modulated across any modality.

Requirements

PhD or equivalent in Bioinformatics, Physics, Chemistry, Computer Science, Structural Biology, or related fields
Experience in protein structural or sequence analysis
Experience in one or more programming languages (e.g. Python)
Experience with training or applying multimodal input (sequence, structure, genetic, small/large molecular, etc.) and output (imaging, omics, etc.) ML models

Nice To Haves

Experience developing or applying modern ML architectures for molecular design models (LLMs, diffusion models, flow-matching, Bayesian Optimization, GNNs, etc.)
Experience with the design of multiple therapeutic modalities
Experience designing de novo binders for specified targets and epitopes to answer biological questions
Experience with cloud engineering production-ready robust and scalable scientific workflows
Experience building and deploying agentic workflows
Demonstrated learning agility, and scientific curiosity while maintaining focus on driving greater impact in the face of uncertainty and change
Ability to generate conclusion reports, present data in team meetings and participate in writing of abstracts and publications for the scientific community

Responsibilities

Work to generate, validate, and integrate multimodal generative AIML models for the de novo design and multi-objective optimization of tool and therapeutic molecules, such as miniproteins, antibodies, antigens, peptides, ADCs, and oligonucleotides.
Guide molecular perturbation experiments that validate mechanisms of disease and show reversal of disease phenotypes and signatures.
Build and exploit agent-orchestrated, integrated Design-Make-Test-Analyze cycles with automated experimental platforms, generating quality data at scale needed for project-specific and foundational models.
Identify and advocate for the opportunities afforded by scientific computation and platform automation and driving therapeutic project plans with predictive technologies.
Collaborate with external groups to further develop protein engineering computational methods.
Predict and evaluate potential disease intervention points for their probability of success to be therapeutically modulated across any modality.