Applied Research Engineer

We are seeking an Applied Research Engineer to bridge the gap between cutting-edge research and production systems at Sakana AI. This is a role for someone who reads papers on Monday, implements them on Tuesday, runs experiments at scale on Wednesday, and ships the results to production by Friday. You will be the critical translation layer between our world-class research team and the engineering organization that deploys their ideas to real customers.

Sakana AI's research agenda is genuinely novel: evolutionary model merging, neural architecture search inspired by biological systems, swarm-based optimization, and techniques for efficiently combining the strengths of multiple foundation models. As an Applied Research Engineer, you will take these ideas from proof-of-concept to robust, reproducible systems. You will build evaluation frameworks that rigorously measure model quality, run ablation studies at scale, fine-tune models for specific enterprise use cases, and develop the tooling that makes our research reproducible.

Day to day, you will split your time between reading and implementing recent papers (both internal and external), running large-scale experiments on our GPU clusters, building evaluation pipelines, and collaborating with product engineers to integrate research outputs into customer-facing systems. You will maintain experiment tracking infrastructure and ensure that every result we publish or ship is reproducible.

The ideal candidate combines deep ML knowledge with strong engineering practices. You should be equally comfortable deriving gradient updates on a whiteboard and writing production-quality Python with comprehensive tests. Experience with evolutionary computation, neural architecture search, or model merging techniques is highly valued and will give you a significant head start.

Requirements:

• 5+ years of experience in applied ML, ML engineering, or research engineering roles
• Strong proficiency in PyTorch with deep familiarity with modern LLM and foundation model architectures
• Demonstrated experience fine-tuning, evaluating, and deploying foundation models in production settings
• Ability to read, implement, and extend research papers independently
• Strong software engineering practices: comprehensive testing, code review discipline, clear documentation
• Experience with experiment tracking and reproducibility tools (Weights & Biases, MLflow, or equivalent)
• Publications or significant contributions to open-source ML projects are a strong plus