Principal Research Engineer, Post-Training
Core
Lead technical vision and execution for post-training systems that transform foundation models into aligned, role-playing chat features for millions of users.
Role type
Principal Research Engineer (Post-Training)
Builds
Production-ready alignment algorithms, training objectives, and scalable data pipelines for large language models.
Domain
Generative AI, Large Language Models (LLMs), Reinforcement Learning, Alignment
Deliverable
production ML models
Required skills
Leading technical projects and teams, deep understanding of transformers and alignment methods, delivering applied ML systems in production, designing production-quality ML infrastructure, training and optimizing large-scale models on GPU, leading LLM training initiatives, product experimentation and A/B testing, software engineering for scalable code
Preferred skills
Hands-on experience with open-source models (Mistral, Qwen), cloud-native ML infrastructure (Kubernetes, Docker), publications in top ML conferences
Technologies
Transformers, Reinforcement Learning, Supervised Fine-Tuning, Preference Optimization, Kubernetes, Docker, GPU-based systems
Responsibilities
Define and drive the technical roadmap for mid- and post-training systems, mentor researchers and engineers, lead development of alignment algorithms and optimization techniques, design efficient training and inference systems, architect scalable data pipelines, establish best practices for experimentation and deployment
Seniority
Principal, strategy & mentorship