Member of Technical Staff (Models)

Sourced from zerogtalent.com

At Jarmin, you will own the post-training loop for AI Employees, defining how models are trained for real-world work. Responsibilities include fine-tuning, reward modeling, and distillation to create smaller, faster, and more cost-effective models. We are looking for hands-on experience with SFT, RLHF, and strong ML fundamentals.

• Post-training and fine-tuning
• Turning agent traces and expert feedback into training data
• Reward modeling and graders for non-verifiable outcomes
• Distillation into smaller, faster, cheaper models
• Designing and running training experiments

• Trained or fine-tuned models and shipped the result into a product
• Hands-on experience with SFT, preference optimization, or RLHF
• Ability to read traces and assess whether a metric measures what matters
• Strong software engineering fundamentals alongside ML depth
• Willingness to work in person in San Francisco

• Adapted open-weight models to a specialized domain
• Built training data or evaluation infrastructure