21 hours ago
Remote, United StatesSenior
Base Salary
$150k - $300k/yr
Responsibilities
- Design and build custom RL environments or digital twins for complex enterprise workflows.
- Post-train LLM-based agents on domain-specific tasks using PPO, GRPO, DPO, RLHF, and related methods.
- Build pipelines that convert human-labeled traces into reinforcement learning training data.
- Architect multi-step reasoning agents with tool-calling and closed learning loops.
- Design reward functions, verifiers, and validation frameworks for pre-deployment testing.
- Translate reinforcement learning research into scalable production systems and contribute to publications.
Requirements
- 3+ years of hands-on experience with RL environment design, reward engineering, and policy optimization.
- Experience fine-tuning LLMs with RLHF, DPO, PPO, or similar methods.
- Software engineering experience beyond research, including scalable pipelines and training infrastructure.
- Experience building LLM-based agents, tool-use systems, and multi-step reasoning workflows.
- Strong Python skills and experience with Gymnasium, RLlib, Stable Baselines, and PyTorch, JAX, or TensorFlow.
- MS or PhD in computer science, machine learning, or a related field, or equivalent experience.
- Preferred: publications at NeurIPS, ICML, ICLR, ACL, or similar venues.
- Preferred: enterprise workflow experience in healthcare, finance, logistics, or compliance.
- Preferred: open-source contributions to CleanRL, TRL, veRL, or agent frameworks.
- Preferred: experience with world models, synthetic data generation, simulation, distributed training, and large-scale RL experimentation.
Benefits
- Hybrid or remote work from Palo Alto, California, or Seattle, Washington.
- Opportunity to shape RL, simulation, and enterprise AI products and deploy research in real-world systems.
Categories
About Centific
Centific builds AI data and engineering platforms and services for enterprises, including curated multimodal datasets, industry‑tuned LLMs, and RAG pipelines backed by vector databases. It monetizes through software, data operations, and consulting to design, deploy, and scale GenAI systems, along with QA automation and digital engineering work. Privately held and headquartered in Redmond, Washington, the company raised a $60M Series A in 2025 to expand its AI data foundry and platform capabilities.