Scientist I / Scientist II, Computational Protein Generation

Generate Biomedicines • Somerville, MA

10h • Hybrid

About The Position

We are seeking a creative, motivated Computational Scientist to join our Model-Driven Design team at Generate:Biomedicines. You will join a talented and collaborative group of ML scientists, engineers, and wet-lab scientists dedicated to redefining how medicines are made. This role sits at the intersection of machine learning, structural biology, and therapeutic development, where you will stay at the leading edge of internal and external models for de novo protein design and systematically benchmark, integrate, and apply them within tightly integrated design–build–test–learn cycles to advance our therapeutic pipeline and impact patients' lives.
The ideal candidate combines deep structural intuition with a demonstrated ability to rapidly assess and apply protein design methods and metrics across diverse design problems, thinks in terms of reusable capabilities, benchmarks, and feedback loops across applications, and uses modern generative models and experimental readouts to guide iterative design cycles across modalities. You don't just run existing tools; you understand what's missing from current approaches and are driven to fill those gaps.
This role is based in our Somerville, MA office with flexibility for hybrid work.
Here's how you will contribute:
Model application and optimization: develop, validate, and productionize de novo protein generation protocols and optimization techniques on our experimental platform, using measured data in-the-loop to iteratively refine models across modalities and therapeutic applications.
Define and implement in silico metrics: design, interpret, and implement biophysical and functional metrics for evaluating generated designs, leveraging existing literature, adapting known metrics to new contexts, and performing original research to benchmark and deploy new scoring approaches.
Benchmark foundation models and guide their application: rigorously evaluate new models and tools and provide quantitative conclusions on where they are best applied to generate new therapeutics, including designing systematic internal benchmarks and discovering how to expand model capabilities to prosecute new therapeutic targets in novel ways and to maximize reuse across targets and programs.
Propose new therapeutic strategies: identify and implement solutions to create new therapeutics through mechanisms of action unlocked by de novo tools and modalities.
Partner cross-functionally to drive therapeutic development: work closely with experimental colleagues, biologists, and clinical scientists to define design objectives, interpret experimental readouts, and guide iterative design-build-test-learn cycles that advance programs.
Advance the state of the art: push forward sequence–structure–function understanding with a focus on reusable platform capabilities and model-informed feedback loops.
Integrate agentic tools into workflows: leverage agentic AI tools to rapidly iterate on models, benchmarks, scores, critics, and other analysis tools, accelerating the pace of discovery.
Build production-quality tools: develop robust, production-ready code in a collaborative team setting and present scientific progress in regular research meetings.

Requirements

• PhD in Computational Biology, Biophysics, Computer Science, or a related field, with demonstrated experience in protein design applications.
• 0–2 years of experience applying computational and/or ML methods to protein design, modeling, or prediction.
• Hands-on experience with machine learning and generative modeling for protein design, including familiarity with modern methods such as RFDiffusion, ProteinMPNN, BindCraft, BoltzDesign, or equivalent approaches and how to deploy or evaluate them in practice.
• Strong structural intuition and understanding of protein biophysics with the ability to quickly assess and adapt design methods and metrics to new problems.
• Familiarity with protein therapeutic modalities such as antibodies, mini-proteins, VHHs, peptides, or enzymes, and an eagerness to deepen expertise across these within de novo design workflows.
• Proficiency in Python and scientific computing; comfort working in a production codebase.
• Experience designing, running, or interpreting benchmarks for computational or generative methods and drawing quantitative conclusions about model applicability and limitations.

Nice To Haves

• Experience designing, executing, and interpreting experiments and experimental data (e.g., binding assays, stability measurements, structural characterization) and using those readouts to inform computational design iterations.
• Familiarity with agentic AI tools and their integration into scientific workflows.
• Exposure to structure-based design techniques and computational tools for modeling protein-protein interactions.
• Track record of translating research ideas into working software or reusable platform components used across multiple projects or applications.

Responsibilities

• develop, validate, and productionize de novo protein generation protocols and optimization techniques on our experimental platform, using measured data in-the-loop to iteratively refine models across modalities and therapeutic applications.
• design, interpret, and implement biophysical and functional metrics for evaluating generated designs, leveraging existing literature, adapting known metrics to new contexts, and performing original research to benchmark and deploy new scoring approaches.
• rigorously evaluate new models and tools and provide quantitative conclusions on where they are best applied to generate new therapeutics, including designing systematic internal benchmarks and discovering how to expand model capabilities to prosecute new therapeutic targets in novel ways and to maximize reuse across targets and programs.
• identify and implement solutions to create new therapeutics through mechanisms of action unlocked by de novo tools and modalities.
• work closely with experimental colleagues, biologists, and clinical scientists to define design objectives, interpret experimental readouts, and guide iterative design-build-test-learn cycles that advance programs.
• push forward sequence–structure–function understanding with a focus on reusable platform capabilities and model-informed feedback loops.
• leverage agentic AI tools to rapidly iterate on models, benchmarks, scores, critics, and other analysis tools, accelerating the pace of discovery.
• develop robust, production-ready code in a collaborative team setting and present scientific progress in regular research meetings.

What This Job Offers

• Job Type: Full-time
• Career Level: Entry Level
• Education Level: Ph.D. or professional degree
• Number of Employees: 1-10 employees