Develop and refine ML pipelines for agent behaviors using prompting, fine-tuning, retrieval-augmented generation, and reinforcement learning techniques. Prototype and experiment with novel agent reasoning, multi-step planning, and tool usage for complex, data-heavy domains. Run structured experiments to evaluate agent performance, optimize reasoning and retrieval, and translate findings into production-ready solutions. Build data pipelines and evaluation frameworks that support rapid iteration and deployment of new agent capabilities. Collaborate with orchestration and infra teams to ensure agent models are robust, scalable, and reliable in production.