Machine Research Engineer (RLHF, Multimodal) – San Francisco

Machine Learning Research Engineer

Location: San Francisco (on-site)

Compensation: $200K–$350K + equity

Visa sponsorship available

A stealth AI startup is building technology that teaches models what great feels like — across writing, design, and creative expression.

They’re hiring a Machine Learning Research Engineer to design and run experiments that shape how models learn taste, tone, and quality. You’ll focus on post-training, reward modeling, and multimodal research, collaborating with creative experts and AI labs on frontier model behavior.

What you’ll do

• Run post-training and fine-tuning experiments (RLHF, DPO, SFT).
• Develop benchmarks and evaluations for subjective model outputs.
• Train reward models, classifiers, and verifiers for tone and style.
• Prototype and test multimodal architectures (text, vision, design).
• Work closely with researchers, engineers, and creative collaborators.

You might be a fit if

• 2+ years in ML research or engineering, especially post-training or alignment.
• Hands-on with LLMs or multimodal models (text + image).
• Strong in Python, PyTorch, and experimental research cycles.
• Experience at a startup or small research team.
• Creative, curious, and interested in how AI learns taste and aesthetics.

If you want to help define how AI understands creativity, emotion, and style — this is the kind of research role that sets the standard.