Job Details

🚨 Research Engineer – Experimental ML Systems

📍 San Francisco, CA | Onsite

🧠 Early-stage AI research lab | Revenue-generating

An AI research lab focused on alignment, interpretability, and reinforcement learning is hiring engineers to build experimental systems that study how models generalize, fail, and become misaligned

This is not a traditional ML infrastructure role focused on scaling training runs

The work is highly exploratory: designing synthetic environments, building research tooling from scratch, and running experiments to better understand model behavior

You'll work on:

🧪 Designing synthetic RL environments for studying model behavior under distribution shift

🎯 Building experimental platforms for generalization, robustness, and alignment research

🔍 Studying reward hacking, deceptive behavior, and goal misgeneralization

🛠️ Prototyping novel training setups and evaluation harnesses

📊 Developing benchmarks that measure internal consistency - not just outputs

⚡ Rapidly testing research hypotheses through code and experiments

Example areas include:

🌍 Toy worlds where models develop deceptive or power-seeking strategies

⚠️ Experimental systems for activation-level interventions

🧠 Robustness benchmarks focused on internal reasoning patterns

Strong fits often come from:

✅ Experimental ML or RL research

✅ Security research, adversarial thinking, or red-teaming

✅ Building small research prototypes vs production systems

✅ Strong engineering ability combined with curiosity and creativity

PhD preferred, but the key requirement is the ability to build novel research systems from scratch

This is not:

❌ Scaling frontier model training

❌ Production ML infrastructure

❌ Data engineering or product ML

This role is for people who want to invent new experimental systems - not optimize existing pipelines

Interested? Hit apply & Drop me a message!