Mid-Training AI Data Scientist — Build Foundation Knowledge

Bison; San Francisco (CA); On-site; USD 350,000 - 475,000; Full time

Benefits offered by this job

Visa sponsorship

Health, dental, vision benefits

Unlimited PTO

Paid parental leave

Relocation support

Job summary

Thinking Machines is hiring for a mid-training lead role in San Francisco to shape model capabilities by sourcing and curating data, building robust training pipelines, and measuring improvements across knowledge and behavior. This role blends research and practical engineering, working between pre-training and post-training to drive scalable data-centric AI.

You’ll collaborate with the post-training team to align behaviors with RL, develop data-quality systems, and explore data-driven

Qualifications

• Proficiency in Python and familiarity with deep learning frameworks (e.g., PyTorch, TensorFlow, or JAX).
• Bachelor’s degree or equivalent experience in Computer Science, Machine Learning, Physics, Mathematics, or related discipline.
• Clarity in communication, ability to explain complex technical concepts in writing.

Responsibilities

• Own the data: source, curate, and synthesize data; build pipelines to filter, deduplicate, verify, and rewrite material into training-grade datasets.
• Improve what the model knows: design interventions with targeted corpora and QA over source documents.
• Instill behaviors and set the prior: evaluate mid-training vs post-training impacts and coordinate with RL teams.
• Build the quality pipeline: develop automatic filtering and scoring; calibrate classifiers and LLM-based judges.
• Develop and tune the recipe: iterate on datasets, training stages, annealing schedules, and hyperparameters.
• Iterate on evals: define evaluations, optimize, and ensure they reflect meaningful results.
• Debug and understand: diagnose results and deepen understanding for future problems.
• Scale and explore: measure performance vs data size and explore new data approaches.

Skills

Python proficiency

Deep learning frameworks

Communication clarity

Education

Bachelor’s degree or equivalent in CS/ML/Physics/Math

Tools

PyTorch

TensorFlow

JAX

Job description

Thinking Machines is hiring for a mid-training lead role in San Francisco to shape model capabilities by sourcing and curating data, building robust training pipelines, and measuring improvements across knowledge and behavior. This role blends research and practical engineering, working between pre-training and post-training to drive scalable data-centric AI.

You’ll collaborate with the post-training team to align behaviors with RL, develop data-quality systems, and explore data-driven