Research, Mid Training
Core
Own late-stage mid-training of frontier AI models to shape fundamental reasoning capabilities, including data strategy, quality pipelines, and recipe iteration.
Role type
Senior IC machine-learning researcher (mid-training)
Builds
Foundation models for reasoning, coding, math, and general capabilities
Domain
Artificial Intelligence / Large Language Models
Deliverable
production ML models
Required skills
Python, deep learning frameworks (PyTorch, TensorFlow, JAX), distributed training debugging, data curation and synthesis, quality classifier development, training recipe iteration, evaluation design
Preferred skills
probability and statistics, synthetic data generation, knowledge injection, model-based data quality systems, alignment research
Technologies
PyTorch, TensorFlow, JAX
Responsibilities
Source, curate, and synthesize training data; design and measure knowledge interventions; build automatic filtering and scoring stacks; iterate on training recipes and hyperparameters; develop and optimize evaluation metrics; debug training configurations and scale methodologies
Seniority
Senior, hands-on IC
