Remote ML Evaluation Architect (Research-Driven)

Alignerr Corp.; Greater London; Remote; GBP 55,000 - 138,000; Part time

Job summary

Alignerr is seeking researchers with deep machine learning knowledge to design evaluation challenges that push the limits of today’s AI systems. This is an hourly remote contract offering 10–40 hours per week, with autonomy over your schedule and collaboration with a global team.

You will develop complex problems grounded in your domain expertise, craft rigorous evaluation criteria, and contribute to gold-standard solutions that shape how models are measured and improved.

Qualifications

• Deep ML knowledge and ability to design evaluation problems.
• Ability to identify gaps where general ML knowledge falls short.
• Experience publishing original research is valued.
• Excellent written communication; articulate complex problems clearly.

Responsibilities

• Design complex, original machine learning problems rooted in your domain.
• Craft evaluation tasks that require advanced domain knowledge beyond standard ML pipelines.
• Draw from your research to create problems that challenge state-of-the-art AI.
• Define rigorous problem statements, evaluation criteria, and gold-standard solutions.
• Assess AI-generated ML solutions for correctness, creativity, and methodological soundness.
• Document problem difficulty levels, required domain knowledge, and AI failure modes.
• Collaborate asynchronously with a global team of researchers and engineers.

Skills

ML knowledge

Domain expertise

Research experience

Strong writing

Independent work

Education

MS or PhD

Job description

Alignerr is seeking researchers with deep machine learning knowledge to design evaluation challenges that push the limits of today’s AI systems. This is an hourly remote contract offering 10–40 hours per week, with autonomy over your schedule and collaboration with a global team.

You will develop complex problems grounded in your domain expertise, craft rigorous evaluation criteria, and contribute to gold-standard solutions that shape how models are measured and improved.