Research Scientist, Cloud AI Research

Job description

Develop agent architectures, planning frameworks, and self-improving foundation models to advance autonomous AI systems. Design and deliver vertical agents across core domains (e.g., infrastructure, forecasting, and optimization) targeting operational efficiencies. Build generalizable horizontal agents to accelerate researcher velocity and automate end-to-end business value generation across Google. Partner with product and engineering teams to translate novel agentic prototypes into scalable enterprise solutions and products. Publish breakthroughs at ML conferences, establish benchmarks, and collaborate across Google. Minimum Qualifications: PhD in Computer Science, a related field, or equivalent practical experience. Experience developing autonomous agents, multi-agent workflows, or agentic frameworks (e.g., AutoGen, CrewAI, LangGraph, tool-use/function-calling, or multi-agent RL). Experience designing, pre-training, fine-tuning, or evaluating Large Language Models (LLMs) and multimodal foundation models. Experience writing code in Python, JavaScript, R, Java, or C++. One or more first-author scientific publication submission(s) on agents in machine learning and natural language processing for conferences, journals, or public repositories (e.g., NeurIPS, ICML, ICLR, ACL, EMNLP). Preferred Qualifications: Experience building evaluation benchmarks, test harnesses, or simulation environments for complex reasoning, code generation, or multimodal tasks. Experience in RL, reward modeling, and human/AI preference alignment techniques (e.g., RLHF, RLAIF, DPO). Experience developing test-time/inference-time compute scaling methods, search-guided decoding, or planning algorithms. Experience developing safe and reliable AI systems, including adversarial, hallucination mitigation, alignment guardrails, model interpretability.