Founding Machine Learning Engineer

Founding ML Engineer About Our Client Our client is an early-stage AI research company building models with real creative ability. They're hiring one of their first engineers to own the systems that turn research ideas into models running in production, working hand in hand with their research team. The Role As a founding ML engineer, you'll own the full path from training to serving. You'll run compute and infrastructure, train models with new capabilities, and make every part of the research-to-production loop faster. This is a rare chance to shape the technical foundation of a company at the earliest stage. What You'll Do Run compute and infrastructure end to end , optimizing for training recipes (CPT, SFT, and RL) and inference Train models across architectures, families, and sizes on new data to unlock new capabilities Optimize model serving , tuning both infrastructure and models for high throughput Work side by side with researchers to develop and refine training recipes Stay close to the product team so model capabilities and product direction inform each other What You Bring Experience training models end to end and building new capabilities into a real product, not just a research project Strong instincts for optimizing and debugging models Experience designing evaluations, both during and after training Proficiency with open-source training infrastructure like NeMo or Ray Core contributions in post-training, ideally including RL beyond RLVR in domains without clean, verifiable answers Genuine curiosity about creative fields. This matters: ML alone won't produce a model with real creative ability, so you'll keep learning the creative domains as you go PRIMIS , Inc. Defunct