Foundation Model Researcher, Molecular Science This role sits inside a well-resourced, interdisciplinary research organization applying frontier large language models and machine learning to molecular science and drug discovery. The environment combines foundational AI research, large-scale model development, and direct collaboration with computational and molecular scientists, with computational infrastructure built specifically for research at this scale. The position owns the research and engineering behind large language and multimodal models for scientific problems: architecture, pre-training, post-training, scaling, and the systems that make all of it run efficiently on high-performance compute. The mandate is not to integrate existing tools but to extend what large-scale models can do in science. Candidates who thrive here combine rigorous research instincts with the ability to ship real, working systems. Molecular science or drug discovery background is not required; depth and versatility in machine learning matter more.
What You'll Do:- Research, develop, and scale large language and multimodal models for complex scientific problems
- Design pre-training pipelines and distributed or parallel training systems for large models
- Explore advanced post-training methods including reinforcement learning, contrastive learning, and instruction tuning
- Build multimodal models spanning text, molecular graphs, 3D structures, time-series data, and other scientific modalities
- Optimize large-scale training and inference across high-performance computing infrastructure
- Partner with ML researchers, computational scientists, and domain experts to turn model advances into new capabilities for molecular science and drug discovery
What We're Looking For:- Exceptional background in machine learning, computer science, mathematics, or a related quantitative field
- Deep expertise in large-scale ML systems, LLM architecture and training, and/or multimodal learning
- Strong Python programming and hands-on ML engineering ability
- Experience with distributed training, model scaling, training infrastructure, or high-performance computing
- Strong command of modern pre-training and/or post-training methods
- Demonstrated ability to conduct rigorous research and also build production-quality systems
- Exceptional record of academic, research, technical, or professional achievement
- Ability to work in the New York City office three days per week
- PreferredExposure to molecular science, structural biology, or drug discovery problems (helpful, not required)
- Experience building models over non-text scientific modalities such as graphs, 3D structures, or time series
Compensation$300K to $800K base salary, plus substantial variable compensation and additional incentives. Hybrid schedule with three days per week in the New York City office.
