Drug Discovery Scientist — AI Evaluation Architect

Obsidian; Greater London; On-site; GBP 50,000 - 75,000; Part time

Benefits offered by this job

Claude Max subscription reimbursed

Claude Code access

Browser-based studio access

Job summary

Obsidian is seeking a researcher to design and deploy an evaluation framework for frontier AI models in drug discovery. You will craft realistic problems, assemble data rooms from real sources, and define rigorous grading criteria that resist shortcut solutions.

You will work in a browser-based studio using Claude tools, produce transparent, source-grounded materials, and lead iterative revisions while maintaining high scientific standards.

Qualifications

• Working expertise in mechanistic enzymology, structure-guided peptide or antibody design, fragment-based discovery, or structure-based medicinal chemistry — PhD or equivalent industry track record.
• Ability to pose a defendable problem that cannot be shortcut.
• Ability to judge reasoning quality, not just final answers.

Responsibilities

• Design the challenge — identify the hidden basin and forcing chain that makes one conclusion inevitable.
• Build the world — assemble a data room from real, source-grounded data. Every decisive value traces to a real source.
• Write the grader — a grading document that scores a submitted memo on substance, not style or method.
• Validate it — run verification and calibration before submission.
• Revise it — you own the rework if a reviewer requests changes.

Skills

Enzymology

Structure-guided design

Fragment-based discovery

Medicinal chemistry

Critical reasoning

Education

PhD or equivalent

Tools

Claude Code

Claude Max

Browser-based studio

Job description

Obsidian is seeking a researcher to design and deploy an evaluation framework for frontier AI models in drug discovery. You will craft realistic problems, assemble data rooms from real sources, and define rigorous grading criteria that resist shortcut solutions.

You will work in a browser-based studio using Claude tools, produce transparent, source-grounded materials, and lead iterative revisions while maintaining high scientific standards.