AI/ML Scientist, Agent Platform (P4021)

What the role needs

• AI/ML Scientist, Agent Platform
• Integrate model APIs, tools, and agent orchestration into maintainable workflows for tasks defined by team scientists.
• Connect data ingestion, agent processing, and structured outputs through reliable pipelines, including delivery to Databricks.
• Implement typed output schemas and deterministic validation checks to detect invalid outputs and support scientific review.
• Build evaluation infrastructure for benchmark runs, result comparisons, and performance assessments led by scientists and collaborators.
• Manage storage, compute, tracing, failure recovery, and reproducibility; monitor and improve workflow cost and latency.
• Maintain shared repositories, tests, packaging, technical documentation, and reproducible development environments.
• Partner with data engineers and RevCore on deployment preparation, secure handling of internal data, and operational handoff.
• Communicate progress, limitations, and technical tradeoffs; incorporate feedback and adapt to priorities across scientific projects.
• Degree in computer science, software engineering, data science, or a related field, or equivalent practical experience.
• Strong Python programming and software design skills, including with agentic coding tools, with experience delivering maintainable applications or computational workflows.
• Experience integrating APIs and implementing typed schemas, automated tests, debugging, and dependency management.

From the employer

AI/ML Scientist, Agent Platform The Opportunity: Seeking a hands-on Contract AI/ML Scientist to build reliable AI agent workflows for scientific data extraction and analysis. The primary focus is agent platform and harness engineering, initially supporting agentic structuring of data from CRO reports focused on TCI synthetic chemistry, with flexibility to support additional agent development projects. Working with team scientists and data engineers, this individual contributor will implement defined requirements, build reusable tools, and prepare prototypes for deployment. Responsibilities: • Integrate model APIs, tools, and agent orchestration into maintainable workflows for tasks defined by team scientists. • Connect data ingestion, agent processing, and structured outputs through reliable pipelines, including delivery to Databricks. • Implement typed output schemas and deterministic validation checks to detect invalid outputs and support scientific review. • Build evaluation infrastructure for benchmark runs, result comparisons, and performance assessments led by scientists and collaborators. • Manage storage, compute, tracing, failure recovery, and reproducibility; monitor and improve workflow cost and latency. • Maintain shared repositories, tests, packaging, technical documentation, and reproducible development environments. • Partner with data engineers and RevCore on deployment preparation, secure handling of internal data, and operational handoff. • Communicate progress, limitations, and technical tradeoffs; incorporate feedback and adapt to priorities across scientific projects. Required Skills, Experience and Education: • Degree in computer science, software engineering, data science, or a related field, or equivalent practical experience. • Strong Python programming and software design skills, including with agentic coding tools, with experience delivering maintainable applications or computational workflows. • Experience integrating APIs and implementing typed schemas, automated tests, debugging, and dependency management. • Experience with Git-based development, code review, packaging, and clear technical documentation. • Experience building data pipelines and reliable workflows, including logging, failure recovery, and reproducibility. • Basic fluency in machine learning and data quality, and sufficient chemistry and biology understanding to follow scientific requirements. • Ability to communicate uncertainty and technical concepts clearly to colleagues with different scientific and software backgrounds. • Ability to independently execute defined work, meet agreed milestones, incorporate feedback, and handle internal data securely. Preferred Skills: • Experience developing LLM applications or agent workflows using the Agents SDK or comparable orchestration frameworks. • Experience with Databricks, cloud platforms, or high-performance computing environments. • Experience with tracing, observability, rate-limit handling, and cost or latency optimization for model APIs. • Experience building evaluation tooling, structured-output validation, or scientific-data applications. • Experience using AI-assisted coding tools with code review and testing or delivering projects in biotechnology or consulting. • Experience with synthetic and/or medicinal chemistry a plus.