Job Details

Job Title: Research Engineer – AI Post-Training, Safety & Alignment

Location: San Francisco, Bay Area

About Us

We're an independent research team working at the cutting edge of AI safety and post-training technique development. Our mission is to make frontier models more reliable, interpretable, and aligned with human values — and we're looking for experienced research engineers to help us push that frontier.

The Role

You'll design and run experiments on the post-training stack — from RLHF and preference optimization to scalable oversight and interpretability. We're particularly interested in engineers who can bridge the gap between theoretical alignment research and robust, production-level experimentation.

Responsibilities

• Implement and evaluate novel post-training techniques (RLHF, DPO, constitutional methods, etc.)
• Collaborate with alignment researchers on safety-critical model evaluations
• Develop tooling, datasets, and metrics for measuring model behavior and robustness
• Contribute to open-source or internal research efforts with publishable outcomes

You Might Be a Fit If You:

• Have experience at top AI labs (e.g., Anthropic, DeepMind, OpenAI, or similar research orgs)
• Are fluent in modern deep learning frameworks (PyTorch, JAX, or equivalent)
• Have hands-on experience with large-scale model training or fine-tuning pipelines
• Care deeply about building systems that do what we intend

Why Join Us

• Work alongside top minds in alignment and post-training research
• Shape the direction of safe, aligned AI development
• Flexible setup (remote, hybrid, or co-location)
• Competitive compensation and meaningful ownership

Apply for a confidential chat!