Research Scientist / Engineer, Open-Ended Discovery

Snapshot

Join an ambitious project focused on Open-Ended Discovery. Our objective is to design and develop methodologies for the autonomous creation of novel artifacts, including new insights, knowledge, algorithms, capabilities, and emergent technologies. We aim to build self-improving systems where the learning process itself generates an endless stream of novel challenges, continually pushing and expanding the model’s or agent’s capabilities. We believe that devising systems that can kickstart and sustain such open-ended co-evolution between agents and their environments is critical to developing increasingly general intelligence, capable of succeeding in surprising emergent scenarios and exhibiting strong out-of-distribution generalization. We believe that combining frontier models such as large language models (LLMs) with open-ended learning approaches is on the critical path to building artificial general intelligence.

About us

Artificial Intelligence could be one of humanity’s most useful inventions. At Google DeepMind, we’re a team of scientists, engineers, machine learning experts and more, working together to advance the state of the art in artificial intelligence. We use our technologies for widespread public benefit and scientific discovery, and collaborate with others on critical challenges, ensuring safety and ethics are the highest priority.

The Role

Key responsibilities: Implement core infrastructure and conduct research to devise systems that kickstart and sustain open-ended discovery and self-improvement. Solve essential problems to generate an endless stream of problems that continually challenge and push the capabilities of participating agents. Develop metrics and scaling laws for generality and emergent intelligence, curate and synthesize diverse learning challenges, and study the mechanisms that drive self-improving, open-ended co-evolution between models/agents and their tasks/environments. Embrace the bitter lesson and seek simple methods that scale, with emphasis on strong systems and infrastructure.

Areas of focus:

• Systems for training agents in complex, evolving, and open-ended environments.


• Infrastructure for generating, curating, and evaluating diverse and novel learning challenges for LLMs, LLM agents, and beyond.


• Methods for efficient, continual learning and adaptation in dynamic and unbounded settings.


• Integrating foundation models (such as LLMs) and open-ended approaches (e.g. evolutionary search or quality-diversity to name a few) into open-ended discovery pipelines.


• Developing methods for never-ending learning in open-ended loops.


• Quantitative evaluations for assessing generality, novelty, feasibility, creativity, emergence, and out-of-distribution generalization in open-ended systems.


• Scaling law science for open-ended discovery and emergent capabilities.

About you

We seek individuals who are passionate about open-ended discovery and believe that continuous, self-generated challenges are crucial for developing truly general intelligence. We strive for simple methods that scale and look for candidates excited to improve models through robust infrastructure, innovative data generation, rigorous evaluations, and efficient compute.

In order to set you up for success as a Research Scientist at Google DeepMind, we look for the following skills and experience:

• MSc or PhD in computer science or machine learning, or equivalent industry experience.


• Experience with prompting, evaluating, and fine-tuning LLMs, building LLM agents, and/or designing and implementing open-ended learning approaches.


• Track record of releases, publications, and/or open source projects relating to open-ended discovery, LLMs, or LLM agents.


• Strong systems and engineering skills in deep learning frameworks like JAX or PyTorch.

In addition, the following would be an advantage:

• Experience building training codebases for LLMs, LLM/RL agents, open-ended methods in complex, evolving environments.


• Expertise optimizing efficiency of distributed training systems and/or inference systems for long-running learning processes.

Application deadline: 1st August

At Google DeepMind, we value diversity of experience, knowledge, backgrounds and perspectives and harness these qualities to create extraordinary impact. We are committed to equal employment opportunity regardless of sex, race, religion or belief, ethnic or national origin, disability, age, citizenship, marital, domestic or civil partnership status, sexual orientation, gender identity, pregnancy, or related condition (including breastfeeding) or any other basis as protected by applicable law. If you have a disability or additional need that requires accommodation, please do not hesitate to let us know.