ML Research Engineer
Core
Train, post-train, and evaluate large language models (LLMs) to build safety, reliability, and optimization layers for AI systems, focusing on policy enforcement and data analysis.
Role type
ML Research Engineer (LLM Training & Safety)
Builds
LLMs, scalable representation pipelines, self-serve analytics tools, and safety/policy systems
Domain
AI Safety, Large Language Models, NLP
Deliverable
production ML models | product features
Required skills
Python, SQL, distributed multi-GPU training, SFT/RLHF/DPO alignment, reward modeling, embeddings, clustering, topic modeling, semantic search, human-in-the-loop labeling, drift monitoring
Preferred skills
Open-source model training, RL methods beyond standard RLHF (e.g., GRPO), moderation/safety model experience, multilingual model training
Technologies
HuggingFace, distributed training frameworks, large-scale data storage/querying systems
Responsibilities
Build scalable representation pipelines for unstructured text; use LLMs for labeling, classification, and data enrichment; deliver insights to change operational decisions; ship self-serve analytics datasets and dashboards; partner with engineering to align pipelines with production constraints
Seniority
Mid-to-Senior, hands-on IC
