Principal Scientist, Evolutionary Search & Autonomous Discovery

Flagship Pioneering•Cambridge, MA

About The Position

Pioneering Intelligence is building an autonomous system capable of exploring large scientific and computational search spaces. A central challenge in this effort is designing algorithms that can generate, evaluate, refine, and recombine candidate solutions over many iterative cycles, continuously improving their performance while maintaining exploration of novel directions. We are looking for a Senior Principal Scientist to invent and build the next generation of evolutionary algorithms. In this role you will develop methods that enable populations of models, hypotheses, programs, or scientific representations to evolve over time through iterative search, selection, variation, and evaluation. Your work will draw on ideas from evolutionary computation, Bayesian inference, reinforcement learning, optimization, and automated scientific discovery. You will help define how AI systems navigate vast spaces of possible solutions — balancing exploration and exploitation, preserving diversity, and discovering novel approaches that would be difficult to identify through conventional optimization alone. This is a senior individual-contributor role with significant scientific latitude and direct influence on the technical direction of PI's autonomous-science platform.

Requirements

PhD in a quantitative field (statistics, computer science, applied mathematics, computational biology, physics, or related) with a strong research track record, or equivalent demonstrated expertise with 5+ years of experience post PhD.
Deep expertise in advanced Bayesian causal modeling and/or evolutionary programming — e.g., probabilistic programming, Bayesian model comparison and weighting, genetic/evolutionary algorithms, quality-diversity methods, or population-based search.
Demonstrated ability to design novel algorithms from first principles and bring them to working, validated implementations.
Strong software engineering skills and comfort running computational experiments at scale (parallel/distributed execution of many model evaluations).
A record of scientific leadership recognized by the field — for example, first-author work at top-tier venues (e.g., NeurIPS, ICML, ICLR), influential publications, or widely adopted open-source tools.
Ability to operate as a senior IC: setting technical direction, collaborating across research and engineering, and mentoring junior colleagues.

Nice To Haves

Experience applying these methods to biological, scientific, or other mechanistic-modeling domains is highly desirable.

Responsibilities

Design and implement algorithms for autonomously generating and evolving populations of candidate solutions at scale.
Develop principled methods for maintaining population diversity, ensuring the search explores genuinely distinct candidates rather than collapsing onto cosmetic variants of the same solution.
Build Bayesian and probabilistic frameworks for evaluating, comparing, and ranking candidates against heterogeneous evidence, including measures of uncertainty and degeneracy across the population.
Define and validate quantitative metrics of diversity and quality across a population of candidates.
Partner with research and engineering teams to integrate these algorithms into closed-loop autonomous science pipelines.
Establish benchmarks and ablations that demonstrate when and why evolved populations outperform single-shot or non-iterative baselines.
Contribute to the broader scientific direction of the team through internal readouts, reusable methods, datasets, prototypes, and external publications.