ML Research Engineer

White Circle is an AI Safety startup looking for ML Research Engineers to train, post-train, and evaluate LLMs. Work includes large-scale data pipelines, alignment techniques (SFT/RLHF/DPO), reward models, and distributed multi-GPU training. The role is hands-on and focused on building scalable systems for AI safety.

• Turn petabytes of unstructured text into structured views; build scalable representation pipelines; use LLMs for labeling and classification; deliver insights and ship self-serve analytics; partner closely with engineering and research.

• Strong Python + SQL; applied NLP/ML experience on real-world text; comfortable at scale with distributed processing; experience with evaluation metrics and reproducibility; prior work with safety/moderation datasets.

• Open-source contributions on HuggingFace/GitHub; experience training models at frontier labs; RL methods beyond RLHF (e.g., GRPO); moderation/classification at scale; multilingual model training.