Research Engineer (AI)

TL;DR

Research Engineer (AI): Training frontier LLMs for scientific discovery with an accent on synthetic data generation, distillation, and continual learning at scale. Focus on scaling compute-efficient model training on thousands of GPUs and developing evaluation methods that correlate with scientific task performance.

Location: Must be based in the US (Menlo Park office available)

Company

An AI and physical sciences lab building state-of-the-art models for scientific research.

What you will do

• Train frontier models to serve as foundations for reinforcement learning.
• Develop methods for synthetic data generation and distillation at scale.
• Collaborate with researchers and physicists to create and guide scientific data curation.
• Optimize compute-efficient LLM training workflows on thousands of GPUs.
• Build high-performance internal tools to investigate data-intelligence relationships.

Requirements

• Experience training LLMs on massive datasets comprising trillions of tokens.
• Ability to calculate scaling laws and compute-optimal hyperparameters.
• Proficiency in generating large-scale, high-quality synthetic data.
• Experience building evaluation systems that correlate with downstream task performance.
• Must have authorization to work in the United States.

Culture & Benefits

• Work in a well-funded, rapidly growing laboratory environment.
• Direct ownership and autonomy over projects without bureaucratic overhead.
• Opportunity to solve complex problems at the intersection of AI and physical science.
• Collaborative environment involving experts in ML, physics, and chemistry.