AI Experimental Systems Research Scientist (Causal Learning amp Adaptive Experimentation)

📋 Description

• • Embark on a groundbreaking journey as an AI Experimental Systems Research Scientist at 3M, a role that redefines the boundaries of machine learning and adaptive systems. Within 3M's esteemed Corporate R&D organization, you will be an integral part of a highly specialized, deeply technical team dedicated to pioneering foundational methodologies for 'always-on' learning systems. These systems are engineered to possess sophisticated reasoning capabilities, actively experiment, and dynamically adapt within complex, ever-changing environments. This unique position is driven by the fundamental understanding that learning systems which fail to actively preserve identifiability, causal validity, and epistemic calibration are inherently flawed, not merely in their performance, but in their very design principles.
• • Your core mission will involve a profound collaboration with a diverse group of researchers spanning the fields of statistics, cognitive science, and machine learning. Together, you will architect and develop sophisticated systems where experimentation, inference, and the quantification of uncertainty are not afterthoughts, but are intrinsically woven into the fabric of the learning process itself. This is a distinct departure from conventional data science or applied machine learning roles, demanding a focus on the intrinsic mechanisms by which learning systems must structure their own experiments, meticulously manage interference and delayed effects, govern their internal representations, and crucially, maintain epistemic correctness over extended periods.
• • This role is exceptionally well-suited for individuals who thrive on working from first principles, possess a passion for designing rigorous experimental machinery, and excel at translating abstract statistical theory into tangible systems that operate with unwavering reliability in real-world, dynamic settings. Your contributions will be pivotal in:
• • Designing and implementing advanced adaptive experimental systems engineered to operate continuously and effectively under conditions of non-stationarity, significant interference, and the complexities of delayed or indirect outcomes.
• • Developing sophisticated causal estimands, innovative randomization schemes, and robust inference procedures whose paramount objectives are identifiability and validity, transcending mere reward optimization.
• • Embedding rigorous experimental control directly into the core architecture of learning systems, including the critical aspect of experimenting on the system's own learning mechanisms, its parameters, and its fundamental representational choices.
• • Translating profound principles derived from experimental design, causal inference, and sequential decision-making into resilient, always-on system behaviors that can be trusted in critical applications.
• • Engaging in a dynamic research process that begins at whiteboards, evolves through in-depth research discussions, and adapts to continuously evolving specifications, rather than being constrained by fixed product requirements or static datasets.
• • Implementing and meticulously maintaining research code that robustly supports hierarchical experimentation, sophisticated baseline control streams, and statistically valid online inference, ensuring the integrity of the learning process.
• • Creating essential diagnostics, advanced monitoring tools, and critical guardrails designed to ensure that learning systems remain accurately calibrated and do not inadvertently stabilize spurious or misleading structures over time.
• • Collaborating closely with interdisciplinary researchers to rigorously stress-test experimental learning mechanisms under realistic, and potentially adversarial, conditions, pushing the boundaries of system resilience and adaptability.
• • This is an opportunity to shape the future of AI by building systems that learn not just from data, but from intelligently designed experiments, ensuring robust, reliable, and ethically sound artificial intelligence.