Senior AI Machine Learning Engineer

The client is seeking a Senior AI/ML Engineer focused on foundation models and agentic AI for life sciences. The role will develop and train proprietary foundation models using multi-omics data, design algorithms to integrate and differentiate across omics modalities, and own end-to-end training workflows including data ingestion, tokenization, pretraining, and evaluation. The position also emphasizes advancing state-of-the-art capabilities in building models that do not yet exist. In addition to model development, the engineer will build and deploy AI agents to support researchers across therapeutic development, computational chemistry, and biology. Responsibilities may include designing multi-agent architectures, integrating agents with internal databases and experimental systems as well as third-party scientific tools, and creating agentic workflows for applications such as cosmetics RandD, drug discovery, and clinical analysis. The engineer will also oversee production ML infrastructure by handling model deployment, scaling, and serving, and by optimizing inference and ensuring reliability, performance, and observability for deployed models. The required qualifications include strong, hands-on ML/AI engineering experience with real code, deep proficiency in Python and PyTorch, and experience training, fine-tuning, and deploying foundation models from scratch or from pretrained checkpoints. The engineer should be familiar with the Hugging Face ecosystem and have experience building agentic pipelines with LangChain and/or LangGraph, along with working knowledge of NVIDIA tooling for inference and orchestration. Comfort with large-scale distributed training tools, strong understanding of modern model architectures, and the ability to collaborate across bio, chem, software, and clinical teams are expected, including clear communication skills; preferred qualifications include experience with omics data, computational biology/chemistry/bioinformatics background, and prior work on production multi-agent systems. Confidential