Principal AI Scientist

About the role

This position is for a Principal AI Scientist to help launch a new AI research program within Polytope Bio, a residency project at Astera. You will be instrumental in shaping the scientific direction and modeling approaches for a system that connects advanced AI models with high-throughput biological experiments. The role focuses on creating generative biology applications by aligning AI models with real-world biological data. You will define the technical roadmap and ensure that scientific hypotheses are translated into executable machine learning workflows. The position requires deep collaboration between computational research and wet-lab experimentation to validate model predictions. You will own the end-to-end process from hypothesis generation to experimental validation in a fast-paced environment. This is a foundational role that will set the standards for future biological AI systems at Astera.

Key facts

What you'll do

• Collaborate with the founder to define the scientific vision, research direction, and technical execution of the program.
• Design and implement post-training methods that convert biological measurements into reward signals for language models.
• Work hands-on to build and debug model training infrastructure, models, training loops, and evaluation metrics.
• Partner with the experimental team to integrate lab measurements with model learning into a unified system.
• Translate complex biological constraints into formal objectives for AI model optimization and training.
• Lead the design of data pipelines that ingest heterogeneous experimental readouts for model consumption.
• Architect scalable training frameworks that leverage large GPU clusters for iterative model development.
• Establish evaluation benchmarks that link model performance to meaningful biological outcomes.
• Mentor junior researchers by providing technical guidance on model architecture and training strategies.
• Drive the publication of research findings through papers, open-source contributions, and internal demos.
• Identify gaps in existing biological datasets and propose novel data collection strategies with the lab.
• Implement robust monitoring of model behavior to ensure scientific validity and reproducibility.
• Explore emergent capabilities in generative biology models through controlled experimentation.
• Coordinate cross-functional discussions to align technical milestones with biological feasibility.

Requirements

• PhD in machine learning, computational biology, or a related field, with 2 to 5 years of industry research experience post-PhD. Accomplished researchers without a PhD are also encouraged to apply.
• Proven experience training models from scratch, including debugging and owning training runs.
• Hands-on experience with generative diffusion models or transformer architectures.
• In-depth understanding of modern reinforcement learning and preference-optimization methods for deep learning.
• A track record of strong research contributions through publications, open-source projects, or deployed models.
• Ability to operate with ambiguity and take ownership of research, with an interest in building infrastructure and growing a team.
• Demonstrated ability to write clean, maintainable, and well-documented code for production systems.
• Strong analytical skills to interpret complex experimental results and adjust modeling approaches accordingly.

Nice to have

• Familiarity with biological research, such as protein modeling, sequence models, or structural biology.
• Experience building and scaling training infrastructure on large GPU clusters.
• Prior team management and technical leadership experience.

Practical notes

This role offers a comprehensive benefits package, including health, vision, dental, and a company-sponsored retirement plan. The project has significant financial runway and dedicated GPU resources. There is potential for a co-founding leadership role in a future spinout, contingent on project success. The position is based in New York or San Francisco, with remote flexibility subject to operational needs. The engagement is full-time with expectations for deep collaboration during core business hours. Travel to Astera facilities may be required periodically to align with experimental milestones. The start date will be determined based on the successful completion of onboarding requirements and resource availability. Applicants must be authorized to work in the United States without sponsorship for this position.