Mid-Training Model Engineer — ML Data & Scaling

Maven Ventures; Palo Alto (CA); On-site; USD 180,000 - 440,000; Full time

Benefits offered by this job

Equity

Medical, Vision, Dental

401(k) Retirement Plan

Disability Insurance

Life Insurance

Employee discounts

Job summary

SpaceXSI is seeking a highly skilled ML engineer to help scale synthetic data pipelines and distill model intelligence into compact flash models. You will design and optimize mid-training data recipes and evaluations to push the capabilities of flagship models.

Proficiency in Python, PyTorch, JAX, Rust, or C++ and experience with Spark/Ray is required. You will contribute to a robust data processing stack and advanced experimentation across modalities.

Qualifications

• Proficiency writing production-grade code in Python, PyTorch, JAX, Rust or C++.
• Experience with large-scale data processing frameworks (Spark, Ray).
• Ability to design and run ML experiments and data pipelines.

Responsibilities

• Scale synthetic coding data to trillions of tokens with large-scale docker verification.
• Distill the intelligence of flagship models into compact flash models via synthetic data generation.
• Optimize mid-training data mixtures to boost RL ceilings.
• Engineer long-context data recipes for training pipelines.
• Develop robust and diverse evaluation for mid-training checkpoints.

Skills

Python

PyTorch

JAX

Rust

C++

Spark

Ray

Tools

Spark

Ray

Job description

SpaceXSI is seeking a highly skilled ML engineer to help scale synthetic data pipelines and distill model intelligence into compact flash models. You will design and optimize mid-training data recipes and evaluations to push the capabilities of flagship models.

Proficiency in Python, PyTorch, JAX, Rust, or C++ and experience with Spark/Ray is required. You will contribute to a robust data processing stack and advanced experimentation across modalities.