Principle LLM Research Scientist

About the roleA5 Labs is redefining the boundaries of AI-driven security in competitive online environments. We specialize in ensuring fair play, integrity, and trust in high-stakes, strategy-based games such as online poker and real-time competitive gaming platforms.We're seeking an experienced researcher who use large language models (LLMs) for reasoning & planning, integrate with external tools/search engines/solvers to solve game‑math or strategy problems, build pipelines for SFT/RLHF/RLAIF, apply to online gaming/poker/strategy‑game domain.Seniority: 5‑10+ years industry experience (with research or applied ML/AI) or earlier with PhD with strong publications.Domain Fit: Experience either in game AI (multiplayer/real‑time/strategy) and/or LLM/tool‑use/agent research.Must-Have Expertise (Not Optional)Published research (papers, conferences) on LLMs, planning agents, RL, tool‑use architecturesWorked in research labs or advanced product teams (AI, game studios, large tech)Experience bridging research to product (i.e., taking LLM / agent models into production or product prototype)Key Skills & Competencies requirements:Core Tech Skills:LLM workflows: SFT , RLHF , RLAIF or equivalent.Tool‑use/agent architecture: Using LLMs as planners/planners & search, integrating external tools (e.g., search engines, heuristic solvers, game simulators).Game maths/strategy/search/planning: Knowledge of search algorithms (e.g., minimax, MCTS), planning, multi‑agent reasoning, perhaps game theory.Programming & experimentation: Python (PyTorch/TensorFlow), experience fine‑tuning large models, designing experiments, feature engineering for game or agent use‑cases.Pipeline & data engineering: Building data pipelines for training/fine‑tuning (label generation, RL loops), data infrastructure (large datasets, multi‑modal gameplay/device/network logs).Real‑time/large‑scale/production readiness: Awareness of latency constraints, gaming backend environments, high‐concurrency systems.Evaluation & metrics: Strong understanding of evaluation metrics (precision/recall/F1/AUC) especially in detection contexts (anti‐cheat or strategy agent contexts).Domain: Bonus if familiar with gaming ecosystems (online poker/strategy games/multiplayer), real gameplay logs, cheat detection or fairness in games.Research mindset: Ability to publish or present work; strong in mathematics/statistics, algorithm design, and literate in academic research/literature.Cross‑functional communication: Able to work with product, engineering, ops, and translate research outputs into deployed features.Preferred Skills:Experience with reinforcement learning or planning agents (RL, hierarchical RL, multi‑agent RL)Familiarity with solver outputs, equilibrium strategies (especially relevant if targeting poker domain)Experience with graph databases/network analysis (for collusion detection or adversarial networks)Cloud training infrastructure (distributed training, data lakes, large compute)Publications/contributions in AI/game research, open source contributions.Education & BackgroundMinimum Education:MS in Computer Science / Machine Learning / Artificial Intelligence ORPhD (strong preference) in CS/AI/ML, Game Theory, Reinforcement Learning, Planning, or related.