Member of Technical Staff, Post-Training

About the Company

Radical Numerics is an AI research lab building general biological intelligence. Their mission is to master the code of life to reduce human suffering, combining distributed systems, model architecture, and numerics research with biological challenges.

Responsibilities

• Develop and tune post-training recipes, including datasets, reward signals, and hyperparameters for biological world models.
• Build and refine evaluation suites for biological reasoning, scientific usefulness, and model reliability.
• Debug model behavior end-to-end, investigating failure modes in training runs and model outputs.
• Explore preference- and feedback-driven learning methods such as reward modeling and synthetic feedback.
• Curate and generate high-quality post-training datasets grounded in biological workflows.

Requirements

• Strong track record in ML research or engineering, specifically in frontier-model training or alignment.
• Proficiency in Python and PyTorch for building production-quality research infrastructure.
• Ability to design careful experiments and interpret ambiguous results in large-scale workflows.

Preferred Qualifications

• Experience with RLHF, RLAIF, or preference optimization.
• Experience designing evaluation frameworks for model safety or scientific performance.
• Background in applied math, systems, or computational biology.