Job Details

Job Title: Research Engineer – Foundation Model Training & Scaling

Location: San Francisco, CA

About Us

We're a tight-knit group of engineers and researchers in San Francisco building frontier-scale models from the ground up. We believe that deep understanding starts before fine-tuning — in how models are trained, what data they see, and how training dynamics evolve. We're obsessed with infrastructure, data curation, and the scaling laws that shape intelligence. Our culture blends hands-on experimentation with scientific clarity and a builder's bias toward action.

The Role

You'll take ownership of large-scale pre-training pipelines: data, compute, and model optimization. That means shaping data mixtures, tuning distributed training systems, and discovering new scaling behaviors. Your work will enable the next generation of aligned and powerful models — robust from the first training step.

Responsibilities

• Run and iterate large-scale pre-training experiments
• Design scalable systems for distributed compute and efficient checkpointing
• Collaborate with research scientists on training dynamics and architecture design
• Analyze performance and scaling trends, turning insights into new experiments
• Maintain a clean, reproducible experimental stack with strong metrics discipline

You'll Thrive Here If You:

• Have experience training or fine-tuning large language or vision models
• Are fluent in deep learning frameworks like PyTorch or JAX
• Understand the nuances of large-scale data pipelines and optimization
• Enjoy blending research curiosity with operational excellence

What We Offer

• Compute, autonomy, and resources to do real foundational work
• A collaborative, research-driven culture without bureaucracy
• San Francisco HQ with relocation support
• Competitive compensation, meaningful equity, and lasting technical impact

Apply for a confidential chat!