Research Engineer, AGI Safety and Alignment, DeepMind

Research Engineer, AGI Safety and Alignment, DeepMind Overview

The Artificial General Intelligence (AGI) Safety and Alignment Team (ASAT) at DeepMind is dedicated to mitigating existential and catastrophic risks associated with AGI and future Artificial Superintelligence (ASI). This role involves researching innovative techniques and collaborating with various teams within Google to apply these findings effectively. The team also provides guidance to executive leadership regarding safety concerns.

ASAT comprises specialized sub-teams that focus on enhancing the alignment of future AI systems, identifying and rectifying sources of misalignment, and exploring advanced alignment techniques. The team prepares for potential AGI risks by simulating scenarios and utilizing interpretability techniques to address practical challenges such as model forensics and evaluation awareness. Additionally, the team is responsible for developing control mechanisms for agents to safeguard against potential misalignments in internal deployments.

Key responsibilities include:

• Conducting research on new alignment methods, studying alignment failures, and applying AGI-scalable alignment techniques to frontier models.
• Developing adversarially robust AGI control systems and implementing them in production.
• Researching interpretability techniques to understand the cognitive processes of AI systems.
• Collaborating with product teams to ensure that research findings are effectively integrated into practical applications.

To qualify for this position, candidates should possess:

• A Bachelor’s degree in Computer Science, Software Engineering, or a related field, or equivalent practical experience.
• A minimum of three years of experience in software development, machine learning engineering, or machine learning research.
• Experience working with research teams.
• Preferred qualifications include experience in applied research aimed at improving the safety and alignment of advanced AI systems and familiarity with training large models, including supervised fine-tuning and reinforcement learning from human feedback (RLHF).

Joining DeepMind offers the chance to be part of a leading AI lab that prioritizes safety and ethics while striving for public benefit and scientific advancement. The organization values diversity and is committed to creating an inclusive environment for all employees.

Degree Requirement: Bachelor’s

Visa Sponsorship Promising