Staff Research Scientist - Reinforcement Learning

Centific

1d•$200,000 - $250,000

About The Position

Centific is a frontier AI data foundry that curates diverse, high-quality data, using our purpose-built technology platforms to empower the Magnificent Seven and our enterprise clients with safe, scalable AI deployment. Our team includes more than 150 PhDs and data scientists, along with more than 4,000 AI practitioners and engineers. We harness the power of an integrated solution ecosystem—comprising industry-leading partnerships and 1.8 million vertical domain experts in more than 230 markets—to create contextual, multilingual, pre-trained datasets; fine-tuned, industry-specific LLMs; and RAG pipelines supported by vector databases. Our zero-distance innovation™ solutions for GenAI can reduce GenAI costs by up to 80% and bring solutions to market 50% faster. Our mission is to bridge the gap between AI creators and industry leaders by bringing best practices in GenAI to unicorn innovators and enterprise customers. We aim to help these organizations unlock significant business value by deploying GenAI at scale, helping to ensure they stay at the forefront of technological advancement and maintain a competitive edge in their respective markets.

Requirements

7+ years in ML/AI research or engineering; 3+ years at senior/staff level
MS or PhD in Computer Science, Machine Learning, or related field (or equivalent)
5+ years hands-on RL — environment design, reward engineering, policy optimization — with at least one production deployment
3+ years fine-tuning LLMs with hands-on RL post-training (RLHF, DPO, GRPO, PPO)
Expert-level implementation of RLHF pipelines, reward modeling (Bradley-Terry), DPO, and KTO
Working knowledge of modern post-training and rollout-serving libraries (TRL, veRL, OpenRLHF, SkyRL)
Experience building LLM-based agents: tool use, multi-turn reasoning, trajectory evaluation
Strong Python and software engineering skills — comfortable building production pipelines, not just notebooks
Deep expertise in MDPs, policy gradient methods (PPO, SAC), and temporal difference learning
Hands-on experience with Gymnasium-based environments and reward engineering (sparse vs. dense)

Nice To Haves

Publications at NeurIPS, ICML, ICLR, ACL, COLM, or similar venues
Open-source contributions to post-training or agent frameworks (TRL, veRL, OpenRLHF, SkyRL)
Experience with Offline RL (CQL, IQL), Model-based RL / World Models, or Hierarchical RL
Background in synthetic data generation, simulation, or world models
Domain experience in healthcare, finance, logistics, or compliance
Distributed training on GPU clusters

Responsibilities

Design simulation environments and digital twins for enterprise workflows
Post-train LLM agents using RLHF, DPO, GRPO, PPO, and emerging methods
Build pipelines that convert human-labeled traces and verifiable signals into training data
Architect multi-turn, tool-using agents with closed learning loops
Design reward functions and verifiers that resist reward hacking and reflect real task outcomes
Set the technical bar across the team — architecture, code review, engineering standards
Mentor researchers and engineers; drive technical direction through influence
Translate research into production; contribute to publications