Senior Research Scientist, AI Coding Agents, Jules, Labs: Google

This role is with the Jules AI team within Google Labs, an organization focused on incubating early-stage, high-impact efforts. The team's mission is to revolutionize how software is designed, developed, and maintained by building the world's most capable and reliable AI coding agent. As a Research Scientist, you will conduct foundational and applied research to solve fundamental problems in AI, including long-horizon planning and reasoning, code understanding, and self-improving systems.

Responsibilities
Define and pursue a long-term applied research agenda to overcome the fundamental limitations of LLMs in complex software engineering tasks.

Design and implement novel agentic architectures, exploring frontiers like automated workflow optimization and multi-agent decomposition.

Lead research into synthetic data generation and verification methods to create data flywheels for improving Gemini.

Develop high-fidelity evaluation frameworks to measure the agent's impact on software quality and developer productivity.

Integrate principles from adjacent domains like machine learning, NLP, program analysis, and formal methods to address ambiguous tasks.

Requirements
Minimum Qualifications:

PhD degree in Computer Science, a related field, or equivalent practical experience.

2 years of experience leading a research agenda.

Experience in one or more of the following: machine learning, large language models, agentic AI (planning, tool use, memory), reinforcement learning, or statistical learning.

At least one scientific publication submission for a top conference or journal (e.g., NeurIPS, ICML, ICLR).

Preferred Qualifications:

Experience building, training, and post-training large language models (LLMs).

Experience with program analysis, program synthesis, automated program repair, or designing developer tools.

Experience designing and implementing novel evaluation benchmarks for LLMs or agentic systems.

Experience in Python software engineering and modern ML frameworks like JAX and TensorFlow.