Job Description

About Us:

Grafton Biosciences is a stealth-mode, San Francisco-based biotech startup focused on solving disease through groundbreaking innovations in early detection and therapeutics. We are combining breakthroughs in synthetic biology, machine learning, and manufacturing to fundamentally extend healthy human lifespans. We’re looking for passionate team members who want to shape the future.

Role: Generative AI Scientist, Computational Chemistry & Drug Discovery

We are seeking a highly specialized scientist at the intersection of machine learning, chemistry, and biology. Your primary focus will be on adapting and creating novel generative AI models to revolutionize our drug discovery process, with a specific emphasis on molecular docking and binding prediction. The ideal candidate is not just a data scientist, but a computational biology scientist who has hands-on experience modifying the core engines of docking and folding software. You will work in close collaboration with our computational chemistry and biology teams to design novel therapeutics.

Key Responsibilities:

• Develop and modify generative AI-driven molecular docking software to improve prediction accuracy, speed, and exploration of the chemical space.
• Design, train, and validate novel generative models (e.g., diffusion models, GANs, VAEs) specifically for creating therapeutic candidates with desired properties.
• Implement and refine models for protein/nucleic acid structure prediction and interaction analysis, potentially leveraging or extending generative docking architectures like DiffDock and NeuralPlexes.
• Apply and advance equivariant graph machine learning techniques for sophisticated representation of molecular structures and interactions.
• Manage and curate large-scale structural and functional biological datasets for model training and evaluation.
• Collaborate deeply with computational chemists and biologists to integrate ML models with molecular simulation data and translate predictions into actionable experimental hypotheses.
• Stay abreast of cutting-edge research in generative models, geometric deep learning, and their applications in chemistry and biology.

Qualifications:

To address the specific needs of this role, candidates must demonstrate experience in the following core areas. Applications without this experience will not be considered .

• Must-Have: Applied Data Science in a Biological/Chemical Context: Proven experience applying machine learning to complex biological or chemical datasets (e.g., structural biology, proteomics, cheminformatics, high-throughput screening data). You must understand the underlying science and the why behind the data, not just the algorithms.
• Must-Have: Experience with Generative AI for Molecular Docking: Hands-on experience modifying, customizing, or developing generative AI-driven molecular docking software or related simulation tools (e.g., DiffDock, NeuralPlexer, GNINA, or similar). This is a critical requirement and goes far beyond simply using off-the-shelf docking programs.

Essential Qualifications:

• Ph.D. or Master's degree in Computational Biology, Computational Chemistry, Biophysics, Computer Science, or a related field with a strong focus on biological applications.
• Proven experience developing generative models (e.g., diffusion models, VAEs) for biological problems.
• Strong understanding of the principles of docking and molecular interactions.
• Proficiency in Python and deep learning frameworks such as PyTorch (preferred).
• Experience with graph neural networks (GNNs), particularly equivariant architectures.
• Proficiency in Linux environments and Bash scripting.
• Experience with cloud computing platforms (Azure or AWS) for large-scale computational workloads.
• Excellent problem-solving abilities and a strong analytical mindset.
• Effective communication and collaboration skills, with an ability to bridge the gap between machine learning and chemistry.

Preferred Qualifications:

• Big Plus: Expertise in dimensionality reduction techniques (e.g., t-SNE, PCA, Tensor Decomposition techniques) for the analysis and visualization of complex, high-dimensional biological data.
• Experience with reinforcement learning (RL) and its application to scientific discovery.
• Familiarity with classical molecular dynamics (MD) or quantum mechanics (QM) simulation packages.
• A track record of publications in relevant machine learning or computational science venues (e.g., NeurIPS, ICML, ICLR, JCB, PNAS).
• Experience deploying machine learning models into production or research pipelines.

What We Offer:

• Competitive compensation.
• Comprehensive health, dental and vision coverage.
• Opportunity to define a new therapeutic‑design paradigm and see your work progress through the clinic.

Screening Questions

If you are a particularly good fit for this role , please email careers@graftonbio.com with responses to the following questions. The email subject should be: GenAI Scientist - [Your Last Name].

(1) Please describe one project where you personally applied machine-learning methods to a complex biological or chemical dataset (e.g., structural biology, proteomics, HTS, cheminformatics). Be sure to cover:

1. The biological or chemical question you were addressing and why the dataset mattered.
2. The nature, size, and quirks of the data (noise, class imbalance, missing values, etc.).
3. The specific ML/AI approaches you chose, how you validated them, and the biological insight or decision they enabled.

We’re looking for evidence that you understand the science driving the data—not just the algorithms.

(2) Please outline your hands-on experience modifying, customizing, or creating generative AI-based molecular docking software (e.g., DiffDock, NeuralPlexer, GNINA). In bullet form, address the following:

• Which codebase(s) you worked on and your precise contributions (e.g., new loss function, equivariant GNN module, GPU acceleration).
• How your changes altered docking performance, speed, or chemical-space exploration—include quantitative metrics if available.
• Any publications, internal reports, or repositories that demonstrate this work (links welcomed).

This role requires deeper engagement than simply running pre-built docking tools—please highlight modifications you personally implemented.

Job Tags

Similar Jobs