Product Manager, Agentic Systems

Job Overview

Staff Product Manager, Agentic Systems

Recursion

Join to apply for the staff product manager, agentic systems role r work will change lives, including your own. Recursion is leading an era of autonomous science – an adaptable system where AI agents navigate the complexity of biology and chemistry to discover medicines faster and better. As the Staff Product Manager for Agentic Systems, you will define the technical and scientific capabilities required for our next suite of capabilities.

You will sit at the intersection of our massive proprietary data generation engine and our cutting‑edge AI models, building the ‘nervous system’ that allows agents to reason, plan, and execute experiments in our automated labs. This role is not about maintaining a static roadmap; it is about navigating the frontier of a rapidly evolving field.

• Define the Architecture for Autonomy:
  Partner with engineering leadership to scope and build the systems that connect our in silico models (the ‘brain’) with our physical automated labs (the ‘body’), enabling closed‑loop, autonomous discovery.
• Drive Hypothesis‑Driven Product Development:
  Lead cycles of experimentation to test different agentic frameworks. You will embrace ambiguity, helping the team decide when to build durable shared services and when to build rapid, throw‑away prototypes to learn what works.
• Operate as a Translator:
  Bridge the gap between wet‑lab realities and dry‑lab possibilities. You will translate the needs of drug discovery programs into technical requirements for agent reasoning, ensuring our systems optimise for information gain rather than just volume.
• Evangelise the ‘Human‑in‑the‑loop’ Evolution:
  Work with scientific stakeholders to define interfaces where humans review, validate, and shape agent reasoning, ensuring our scientists evolve from operators to architects of discovery.

You will join a cross‑functional team of software engineers, data scientists, and AI/ML scientists who build the technical bedrock that enables autonomous science, including agent orchestration, guardrails, and connectivity between our digital and physical assets. You will collaborate closely with Discovery teams (the users of these agents), AI Research teams (who build cutting‑edge models), and automated biology and chemistry lab teams (who generate the data that feeds into the models).

• 5+ years of product management experience, with a focus on platform, infrastructure, or AI/ML products where the technical solution was not immediately obvious.
• Experimentation‑first mindset: A proven track record of managing products through rapid prototyping cycles, understanding when to build to last versus build to learn.
• Technical fluency in modern AI:
  Fluency in concepts of LLMs, agentic workflows, APIs, and modern data infrastructure; can hold own in a debate about orchestration architectures.
• Systems thinking:
  Ability to visualise complex ecosystems and anticipate downstream impacts of lab protocol changes.
• Communication & influence:
  Strong written and oral skills to align diverse stakeholders (PhDs in Biology, Robotics Engineers, AI Researchers) around a unified technical vision.

This is an office‑based, hybrid position at one of our offices in either Salt Lake City, Utah or New York City, New York. Employees are expected to work in the office at least 50% of the time. Based on the skill and level of experience required for this role, the estimated current annual base range is $141,400 – $204,800. You will also be eligible for an annual bonus and equity compensation, as well as a comprehensive benefits package.

• We act boldly with integrity. We are unconstrained in our thinking, take calculated risks, and push boundaries, but never at the expense of ethics, science, or trust.
• We care deeply and engage directly. Caring means holding a deep sense of responsibility and respect—showing up, speaking honestly, and taking action.
• We learn actively and adapt rapidly. Progress comes from doing. We experiment, test, and refine, embracing iteration over…