Machine Learning Engineer

EPM Scientific is supporting a growing research driven team building advanced AI agents designed to accelerate scientific discovery. This group develops large scale language model based systems, autonomous agents, and the infrastructure required to deploy them in real scientific workflows. Their work brings together modern machine learning, agent architectures, and real world scientific feedback loops, enabling models that directly influence high stakes research decisions.

This role is ideal for someone who wants to work at the intersection of LLM development, agentic AI, and scientific workflow integration, and who wants their engineering and modeling work to have tangible impact in real experimental and analytical settings.

Role Overview
You will design, train, and refine machine learning models and agent systems that support complex scientific tasks. You will collaborate closely with scientists, engineers, and domain experts to understand real workflows, interpret results, and integrate experimental or field derived feedback into iterative model improvements.

Your Work Will Include:

• Training and refining large scale LLM models, ensuring efficient experimentation and scalable training workflows.
• Designing agent architectures that integrate tool use, prompting strategies, and evaluation frameworks for scientific tasks.
• Building and maintaining data pipelines and inference systems that support robust model training and deployment.
• Testing agent behavior in real scientific workflows and using feedback to guide iterative improvements.
• Collaborating with ML researchers, scientists, and engineers to ensure outputs translate into actionable scientific value.

What You Bring:

• 5-10+ years of experience in ML engineering or applied ML work, ideally involving LLM powered systems or agent architectures.
• Strong Python skills, with experience in PyTorch or JAX; familiarity with TypeScript is a plus.
• Experience with prompting, context engineering, evaluation frameworks, or building LLM driven tools or agents.
• Understanding of distributed training, scalable data pipelines, or ML infrastructure for high performance systems.
• Ability to work closely with scientific or R&D partners and adapt models to real world workflow constraints.
• Interest in developing systems that deliver practical scientific impact, not just conceptual models.