$65–105/hr · Mercor · Part time, 40 hours a week
A research chemist who creates synthesis problems and evaluates AI performance on chemistry challenges for model training and assessment.
What you would do
- Write original synthesis or characterization problems based on your own research, including solution pathways and evaluation criteria for assessing answers
- Evaluate whether AI-proposed synthesis routes demonstrate sound disconnection logic, realistic selectivity, and feasible workup or isolation procedures
- Assess spectroscopic data interpretation and structure assignment, identifying where AI reasoning is supported by evidence and where claims exceed what the data supports
- Document assessments with specific reasoning about gaps between paper chemistry and practical execution due to side reactivity, workup inefficiency, or reagent incompatibilities
Who they want
- Graduate degree in chemistry or closely related field with hands-on laboratory or computational chemistry research experience carried out personally
- Clear written communication ability; most evaluation work involves explaining your reasoning and technical judgments in detail
- Comfort with ambiguous or underspecified problems, and willingness to flag unclear instructions or scope boundaries
- Current familiarity with synthesis and spectroscopic data interpretation obtained through recent hands-on work, not prior experience alone
What the interview asks about
1.Problem authorship from personal research
Original problems grounded in your own work carry credibility that textbook or repurposed problems lack; your judgment about difficulty and relevance reflects real research priorities.
For example: “Describe a synthesis or characterization task from your research. What makes it challenging? How would you structure it for an AI to attempt?”
2.Synthesis route feasibility assessment
Evaluating whether a disconnection makes sense in principle requires different expertise than assessing whether it is actually executable; your lab experience bridges this gap.
For example: “An AI proposes a synthesis route using thermodynamically allowed reagents and standard mechanisms. Walk through whether the approach is sound and what's missing.”
3.Spectroscopic data interpretation accuracy
Overconfident or incorrect structure assignment from spectroscopy is common in AI outputs; your judgment about what the data supports versus what is speculative determines evaluation reliability.
For example: “You're evaluating a spectroscopy structure assignment. The AI proposes a mechanism. Describe what makes a spectroscopy argument convincing and how you'd verify it.”
4.Practical chemistry versus paper chemistry gaps
An AI trained only on chemical literature might never learn which theoretically reasonable transformations fail in practice; your tacit knowledge of experimental realities is irreplaceable.
For example: “An AI proposes a Grignard route to an alcohol using an aryl halide and formaldehyde. Explain what's wrong and how you'd guide someone to see the issue.”
5.Clear technical judgment communication
Clients and AI developers can only act on your findings if you explain your reasoning concretely; vague feedback is not actionable.
For example: “Evaluate an AI-generated synthesis. Judge whether selectivity claims hold, what the most likely byproducts are, and what experiment would definitively test the route.”
A task you may get
Author a complete synthesis or characterization problem from your own research, including theoretical background, solution pathway, and criteria for assessing attempts.
How to prepare
- Identify 3-5 significant results from your recent research that could serve as strong problems; for each, consider what makes it a good test of AI reasoning about chemistry
- Prepare specific examples from your hands-on work where a theoretically sound synthesis failed in practice, and clarify what the actual limitation was
- Practice writing technical evaluations of other chemists' work; review how clearly you communicate your reasoning about synthesis feasibility or spectroscopic interpretation
The facts
- Pay
- $65–105/hr
- Hours
- Part time, 40 hours a week
- Where
- Remote
- Field
- Life, Physical, and Social Science
- Role type
- Talent network
- Posted
- 9/15/2026
We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.