Senior Principal Investigator, Physical AI for Autonomous Discovery

Reporting to the Director, Physical AI for Autonomous Discovery, the Senior Principal Investigator will serve as a hands-on scientific and technical leader developing Physical AI capabilities for autonomous laboratory experimentation. This individual-contributor role combines AI, robotics, laboratory automation, and scientific reasoning to design, implement, and validate systems that perceive laboratory state, recommend actions, learn from experimental data, recover from errors, and improve experimental outcomes.

Key Responsibilities

• Lead the hands-on design, development, and validation of AI models that represent instruments, samples, reagents, robotic systems, workflows, and laboratory state.
• Develop and evaluate adaptive experimental optimization methods using active learning, Bayesian optimization, reinforcement learning, and multi-objective optimization.
• Implement capabilities for physical error detection and correction, including anomaly recognition, root-cause assessment, workflow replanning, recovery actions, and appropriate human escalation.
• Build models that generate testable hypotheses for experimental failures, unexpected results, variability, and irreproducibility, and recommend discriminating follow-up experiments.
• Develop agentic and multi-agent capabilities that support experiment planning and execution monitoring across Design-Make-Test-Analyze cycles.
• Translate scientific and operational needs into technical requirements, prototypes, and production-ready capabilities in partnership with automation, robotics, data science, engineering, and discovery scientists.
• Provide technical guidance, establish scientific best practices, and mentor colleagues without formal people-management responsibility.
• Embed calibrated uncertainty, safety constraints, monitoring, traceability, and governance into autonomous decision-making, escalating strategic tradeoffs to the Director as appropriate.

Key Skills

• Physical AI & World Models: Embodied AI, multimodal perception, state estimation, digital twins, and predictive models of laboratory systems.
• Autonomous Experimentation: Active learning, Bayesian optimization, sequential design, reinforcement learning, and multi-objective optimization.
• Error Intelligence: Anomaly detection, fault diagnosis, causal reasoning, hypothesis generation, recovery planning, and uncertainty quantification.
• Robotics & Automation: Laboratory robotics, liquid handling, sensors, machine vision, orchestration, scheduling, and instrument integration.
• Agentic AI: Scientific agents, multi-agent coordination, planning, tool use, memory, guardrails, and human-on-the-loop controls.
• Scientific Software: Python and modern AI/ML frameworks; production model deployment, APIs, data pipelines, observability, and MLOps.

What Success Looks Like

• Prototype and production capabilities measurably reduce unplanned intervention by detecting, diagnosing, and supporting recovery from laboratory execution failures.
• AI-generated hypotheses explain experimental anomalies and guide informative follow-up experiments that improve data quality and learning efficiency.
• Closed-loop optimization methods are validated against clear scientific and operational benchmarks for quality, throughput, reproducibility, and decision-making.
• Technical workstreams progress from proof of concept to reliable deployment through documented designs, reproducible evaluation, and effective cross-functional execution.
• The individual partners closely with the Director and with J&J IT and Data, Data Science & AI, automation, and scientific teams to deliver robust, secure, and governed agentic capabilities.

Qualifications

• PhD in artificial intelligence, machine learning, robotics, computer science, engineering, computational biology, or a related quantitative field, with substantial relevant research or industry experience.
• Demonstrated track record of independently developing and deploying AI systems in physical, scientific, or similarly complex environments.
• Deep expertise in multiple relevant areas, such as world models, reinforcement learning, Bayesian optimization, causal inference, multimodal AI, robotics, or scientific machine learning.
• Advanced software-development skills, including Python, modern machine-learning frameworks, experimental software design, testing, and reproducible deployment practices.
• Ability to define technical workstreams, solve ambiguous scientific problems, and deliver robust solutions with limited day-to-day direction.
• Strong communication, collaboration, and influence skills for working effectively across multidisciplinary teams in a matrix environment.

Preferred Experience

• Self-driving laboratories, laboratory automation, or closed-loop scientific experimentation.
• Vision-language-action models, digital twins, machine vision, robotic control, or multi-agent systems.
• Biologics, drug discovery, life sciences, or another complex experimental domain.
• Deployment of safety-critical or high-reliability AI systems with monitoring and governance controls.