Member of Technical Staff, Post-Training

Radical Numerics•San Francisco, CA

About The Position

Radical Numerics is an AI research lab focused on building general biological intelligence with the mission to master the code of life and reduce human suffering. The team has a history of significant contributions, including the creation of Evo and pioneering the field of generative genomics. Their work has been recognized in publications like Science and presented at TED2025. Evo has been instrumental in developing tools like CRISPR-Cas9 and generating entire genomes from scratch. Evo 2, featured in Nature, is a large, open-source AI project. Radical Numerics applies principles from distributed systems, model architecture, and numerics research to biological challenges, redesigning the foundation model training stack to transform scientific data into generative models. The company acknowledges the dual nature of these advancements, recognizing that the power to design also brings the responsibility to defend against potential misuse, such as AI-generated bioweapons. Radical Numerics was founded to address both aspects.

Requirements

Strong track record in ML research or engineering, especially in frontier-model training, post-training, alignment, evaluation, data quality, or related areas.
Proficiency in building production-quality software and research infrastructure, ideally in Python and PyTorch.
Comfort debugging large-scale training workflows.
Ability to design careful experiments, interpret ambiguous results, and separate real effects from artifacts, bugs, or benchmark overfitting.
Excellent written and verbal communication skills, especially the ability to explain technical findings clearly across research, engineering, and scientific collaborators.
Curiosity, rigor, and a bias toward iteration: you like improving systems by repeatedly tightening the loop between hypotheses, experiments, and insight.

Nice To Haves

Experience with RLHF, RLAIF, preference optimization, reward modeling, rejection sampling, or other post-training methods for large models.
Experience designing or operating evaluation frameworks for model quality, reliability, safety, or scientific task performance.
Familiarity with synthetic data generation, annotation workflows, or expert-in-the-loop data collection.
Background in applied math, systems, computational biology, or another quantitative scientific field.
Contributions to open-source ML systems, model tooling, or research infrastructure.

Responsibilities

Develop and tune post-training recipes for biological world models.
Design and iterate on post-training stages, datasets, reward signals, and hyperparameters.
Study the impact of data mixtures, objective design, curriculum, and training schedules on model behavior.
Collaborate with the science team to develop and refine evaluation suites for biological reasoning, scientific usefulness, long-context behavior, robustness, and model reliability.
Identify when existing benchmarks become uninformative and need replacement.
Investigate failure modes in training runs and model outputs, distinguishing signal from noise.
Trace problems back to data, optimization, evaluation design, or systems issues.
Explore methods such as preference modeling, reward modeling, synthetic feedback, or related post-training approaches.
Define, curate, or generate high-quality post-training datasets, including expert-informed data, synthetic data, and task-specific examples.
Measure how performance changes with dataset size, recipe complexity, compute budget, and model family to guide future scaling and exploration.
Work closely with colleagues in training systems, architecture, and biology-facing research to ensure post-training methods are grounded in large-scale experimentation and downstream scientific use.