Research Engineer - Novel AI Platforms for Multiscale Alignment
Future of Life Organizations1 year ago
Remote, WorldwideSenior
Base Salary
$140k - $210k/yr
Responsibilities
- Research and develop AI systems and platforms for safety research, internal tooling, and community support.
- Design and implement agent execution environments and interaction platforms.
- Develop optimization algorithms for multi-objective and cooperative AI systems.
- Create conflict-resolution and preference-aggregation mechanisms for multi-agent settings.
- Implement testing frameworks and example environments to validate theoretical approaches.
- Build middleware for secure and efficient agent communication.
- Architect reusable software components for AI alignment research.
- Document system architectures, APIs, and implementation details.
- Collaborate on technical publications and research presentations.
- Contribute to practical implementations and theoretical AI alignment research.
- Support evaluation and refinement of research prototypes.
Requirements
- MS or PhD in Computer Science, AI, or a related field, or equivalent experience.
- Experience with dynamic plan recognition, activity recognition, dynamic multicriteria decision making, multi-agent systems, or AI builder platforms.
- Strong programming skills in Python and Java.
- Ability to independently drive technical projects from concept through implementation.
- Familiarity with AI alignment, control, safety, or related research techniques.
- Experience developing middleware, frameworks, or platforms.
- Broad experience in machine learning and AI systems architecture.
- Literacy in semi-formal semantics.
- Strong communication and technical documentation abilities.
- Preferred qualifications include AI alignment or safety publications, multi-objective optimization, contextual pragmatics, constraint satisfaction, novel multi-agent architectures, LLM integration, distributed systems, security and privacy methods for multi-party computation, game theory, social choice theory, metacognitive systems, metareasoning, dynamical cognitive approaches, open-source contributions, and systems managing competing objectives.