RLHF Specialist

About the role

The RLHF Specialist will design and operate reinforcement‑learning‑from‑human‑feedback pipelines to improve large language models. Working remotely, you will collaborate with ML engineers to align model behavior with safety, factual accuracy, and human values.

Key responsibilities

• Generate high‑quality preference data by ranking model responses on helpfulness, honesty, and harmlessness.
• Design multi‑turn prompts that stress‑test reasoning and safety, and write chain‑of‑thought rationales for reward‑model training.
• Collaborate with engineers to analyse failure modes, identify data gaps, and propose data‑driven interventions.
• Develop and iterate annotation strategies, ensuring consistency across a global team.
• Probe models for biases, hallucinations, and vulnerabilities, documenting findings for optimisation.
• Maintain a personal benchmark set, regularly re‑evaluating new model versions against historical performance.

Required profile

• Minimum 2 years of experience in data annotation, model evaluation, computational linguistics, or AI trust & safety.
• Strong proficiency in Python and deep‑learning frameworks such as PyTorch, JAX, or TensorFlow.
• Deep understanding of reinforcement‑learning concepts (PPO, trust regions, reward hacking) applied to language generation.
• Hands‑on experience fine‑tuning open‑source models (e.g., Llama 2/3, Mistral, Gemma) using LoRA/QLoRA.
• Experience with annotation platforms (LabelBox, Scale AI, Snorkel) and cloud ML services (AWS SageMaker, GCP Vertex AI).

Required skills

• Python
• PyTorch
• JAX
• TensorFlow
• Reinforcement Learning (PPO, trust‑region methods)
• LoRA / QLoRA fine‑tuning
• Open‑source LLMs (Llama 2/3, Mistral, Gemma)
• Annotation tools (LabelBox, Scale AI, Snorkel)
• AWS SageMaker
• GCP Vertex AI