Member of Technical Staff, Staff Physicist, Quantum Information and AI

About FirstPrinciples:
FirstPrinciples is a non-profit organization building an autonomous AI Physicist to understand the nature of reality: the underlying structure, governing principles, and fundamental laws of our universe. We're developing an intelligent system that can explore theoretical frameworks, reason across disciplines, and generate novel insights to tackle the deepest unsolved problems in physics. By combining AI, symbolic reasoning, and autonomous research capabilities, we're developing a platform that goes beyond analyzing existing knowledge to actively contribute to physics research. Our goal is to accelerate progress on the questions that have captivated humanity for centuries.

Job Description:
We are looking for a Member of Technical Staff, Staff Physicist to help build an AI Physicist at the frontier of Quantum Information and AI. You will bring postdoc-level rigor in quantum information theory and turn that expertise into training signal, evaluation methods, and research direction for a rapidly evolving scientific system. This is a researcher role at the intersection of AI and physics: you will help invent new benchmarks, metrics, and evaluation methodologies for what it means to do high-quality research in Quantum Information with AI in the loop. You will work closely with research and engineering teams, and your contributions will flow straight into production model improvements and publishable outcomes.

Key Responsibilities:

Scientific Critique and Research Guidance:

• Review and critique model reasoning in quantum information and adjacent theory (entanglement, channels, capacities, quantum error correction, cryptography, algorithms).
• Identify subtle conceptual errors, missing assumptions, invalid proof steps, and “sounds right” failures.
• Provide clear corrections, alternative derivations, and minimal counterexamples that teach the system what good physics looks like.
• Translate domain judgment into actionable research recommendations for model behavior, reasoning style, and tool use.

Expert Feedback for Model Training:

• Create gold-standard demonstrations and reference solutions suitable for training and fine-tuning.
• Provide structured preferences and rankings over candidate model outputs to improve scientific reasoning quality using expert feedback loops (including RLHF-style workflows).
• Define what “better” means for research outputs: correctness, explicit assumptions, uncertainty calibration, reproducibility, and citation discipline.
• Help build a repeatable pipeline that converts expert scientific judgment into scalable training signal.

Benchmarks, Metrics, and Research Evaluation:

• Co-develop new benchmarks for conducting research in Physics and Quantum Information, with an emphasis on measuring real scientific competence rather than surface-level fluency.
• Define metrics that capture proof validity, assumption tracking, unit and dimensional consistency, asymptotic reasoning, correct theorem usage, and the ability to propose falsifiable next steps.
• Build task suites that reflect real research workflows, including literature-grounded problem framing, derivation under constraints, error diagnosis, and hypothesis refinement.
• Partner with ML and engineering teams to implement these benchmarks as automated evaluation gates and continuous monitoring signals.
• Publish or open-source benchmarks, datasets, and baselines where appropriate to advance the broader scientific community.

Evaluation, Reliability, and Continuous Improvement: 

• Design evaluation suites and rubrics that stress-test the model on hard Quantum Information tasks and expose common failure modes.
• Track recurring error patterns and propose interventions (data improvements, prompt and tool changes, training targets, evaluation gates).
• Maintain internal libraries of known failure classes, fixes, and “red flag” signatures that drive iteration speed without sacrificing rigor.

Collaborators Program and Cross Functional Coordination: 

• Work and help us build our Collaborators program, an external group of expert peers acting like a set of reviewers at a pre-eminent journal.
  Coordinate review cycles and incorporate collaborator feedback into training priorities, benchmark design, and evaluation criteria.
  Align external reviewer standards with internal research goals and engineering constraints, ensuring fast iteration while maintaining scientific defensibility.
  Communicate progress and open questions clearly across collaborators