Senior/Principal AI Scientist, Foundational ML, AI for Biology & Translation (AIBT)

Roche • South San Francisco, CA

5d • $168,100 - $373,800

About The Position

The Senior/Principal AI Scientist will join the Foundation Models team (DELTA) within the AI for Biology & Translation (AIBT) department in Genentech Research and Early Development (gRED). The team conducts AI research aimed at real-world impact in drug and target discovery, focusing on large-scale biological foundation models that can reason across biological modalities, guide experiments, and learn from outcomes.

The role advances the next generation of foundation models for target and drug discovery, including research in multimodal generative modeling, representation learning, LLM mid-training and post-training, agentic science, and reinforcement learning. Applications include genomics, perturbation biology, imaging, and multimodal experimental data. The scientist works in a multidisciplinary environment with AI scientists, AI engineers, and computational biologists, and provides technical and scientific leadership by shaping research directions, developing and testing ideas, addressing modeling, data, and engineering challenges at scale, leading collaborative projects, and publishing in leading machine learning and scientific venues.

At the Principal level, the scientist also defines and drives broader research initiatives, establishes technical direction across related projects, and resolves critical modeling, data, and engineering challenges while remaining hands-on in research and development. The position is based in San Francisco. Relocation benefits are not available. The stated salary range is $168,100–$312,300 for Senior AI Scientist and $201,300–$373,800 for Principal AI Scientist; actual pay depends on experience, qualifications, location, and other permitted factors. A discretionary annual bonus may be available based on individual and company performance. The position qualifies for company benefits described via a link in the posting.

Requirements

• Ph.D. in Computer Science, Machine Learning, Computational Biology, or a related quantitative field.
• Proven track record advancing AI models in research and/or industry, with deep expertise in one or more of large-scale representation learning, multimodal generative models, LLMs, AI agents, or reinforcement learning.
• Strong motivation to advance AI for science and translate AI advances into progress in biology and drug discovery.
• Excellent knowledge of deep learning theory and practice, strong research intuition, and ability to turn open scientific questions into rigorous experiments.
• Proven experience developing and delivering innovative machine learning solutions in the relevant areas.
• Excellent Python programming skills, fluency with modern agentic coding environments, and extensive experience with AI frameworks such as PyTorch or JAX.
• Strong grasp of software engineering, data engineering, and MLOps best practices, including scalable data processing, high-performance computing, and reproducible training and evaluation workflows.
• Strong publication record and active contribution to research communities, including top-tier ML venues such as NeurIPS, ICML, and ICLR, and/or a public portfolio demonstrating substantial contributions to relevant AI projects.
• Excellent communication, collaboration, and problem-solving skills.
• At the Principal level, typically 3+ years of post-Ph.D. research experience in relevant AI areas, or an equivalent portfolio demonstrating comparable depth, independence, and impact.
• At the Principal level, a strong track record of driving complex research initiatives with substantial scientific or technical impact, including sound judgment in identifying high-value problems and making consequential technical decisions.

Nice To Haves

• Prior experience applying AI to biology, chemistry, or drug discovery, especially with multimodal experimental data or experimental design.
• Practical experience bridging innovative machine learning methods and applications in target or drug discovery.
• Experience with biological and chemical modalities and tasks such as molecular structures, single-cell/omics data, perturbation biology, and multimodal biological datasets.
• Hands-on experience developing, fine-tuning, and optimizing LLMs and agentic systems.

Responsibilities

• Design and build foundation models for target and drug discovery, emphasizing large-scale representation learning, multimodal mid-training and post-training, reasoning, and reinforcement learning.
• Develop scalable methods to curate, organize, and integrate diverse data modalities—including molecular structures, biological sequences, omics data, biochemical readouts, and text—into high-quality training data and evaluation tasks.
• Develop models and agentic systems that propose experiments, interpret results, and learn from experimental feedback to advance lab-in-the-loop discovery.
• Build training and evaluation environments from biological data to develop and measure scientific reasoning and discovery capabilities.
• Scale state-of-the-art models to massive datasets, addressing distributed training, efficient experimentation, and reliable data pipelines in collaboration with engineering and MLOps teams.
• Publish in top-tier machine learning venues and scientific journals, and present results at internal and external conferences and workshops.
• At the Principal level, define and drive larger research initiatives, set technical direction across related projects, and resolve critical modeling, data, and engineering challenges while remaining hands-on.

Benefits

• A discretionary annual bonus may be available based on individual and company performance.
• The position qualifies for company benefits detailed at a link provided in the posting.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: Ph.D. or professional degree