Subject Matter Expert (SME) Advanced Technical Reasoning (STEM)
Job Overview
We are seeking Subject Matter Experts (SMEs) across engineering and the sciences to design
rigorous, high-complexity problems that test the multi-step reasoning capabilities of advanced
AI systems. You will apply deep domain expertise to author challenging problems and fully
derived, verifiable solutions that serve as ground-truth benchmarks.
The role is open across disciplines, and contributors work strictly within their own field of
specialization.
Key Responsibilities

¢
Design high-complexity, multi-step problems in your specialization that test the
quantitative and analytical reasoning of advanced AI systems.
Author fully derived solutions that serve as verifiable ground-truth references, ending
in an unambiguous final answer.
Ensure each item is original, self-contained, and well-posed, with a single objectively
verifiable answer.
¢
¢
Independently verify every calculation and intermediate result before submission.
Ensure every item is of high complexity, with genuine domain depth.
Problem/Solution Authoring
For each item, you will produce:

1. Problem Statement
A precisely scoped task with a single objective and one definitive, objectively
verifiable answer.
2. Visual Artifact (multimodal items only)
o
o
An authentic, technically accurate figure (such as a drawing, schematic, plot,
or process diagram) from which the essential data must be interpreted.
3. Worked Solution
A fully derived, step-by-step solution that traces the reasoning from problem
formulation (and visual interpretation, where applicable) to the result.
4. Final Answer
The requested result stated concisely, with correct units, significant figures,
and format.
o
o
Creation Requirements
¢
¢
¢
Self-contained: the problem statement supplies all data,



constraints, and boundary
conditions required for a unique solution, with no unstated assumptions.
Well-posed: a single objective and one verifiable answer, with the required units,
significant figures, and specified answer format.
High complexity: multi-step, non-routine reasoning in which the solver must select
and justify the governing principles and method, rather than apply recall or direct
substitution.
¢
Difficulty: target the Hard tier: multiple reasoning steps, multi-concept synthesis, and
multiple decision points (assumptions, method, validation), and difficult enough that
advanced AI models fail to solve it reliably.
¢
¢
¢
¢
¢
Disciplinary depth: resolution must depend on the core theory and methods of the
field, not on generic quantitative reasoning.
Originality: a novel formulation and parameterization that cannot be retrieved
through online search.
Solution rigor: clearly labelled steps, defined symbols and assumptions, consistent
units and dimensional analysis, and no unjustified leaps in logic.
No AI usage: the problem, solution, and final answer must be authored entirely by
you, without generative AI tools or AI-generated content at any stage of QSA creation.
LaTeX: all equations, symbols, units, and mathematical expressions in the problem,
solution, and final answer must be typeset in proper LaTeX (math mode), with no

plain-text or ASCII math.
¢
¢
Final answer: restricted to the requested result,



with mathematical expressions
typeset in LaTeX.
Figure integrity: authentic, technically accurate, high-resolution and legible, with
labelled axes, scales, or dimensions where relevant, and free of watermarks, logos, or
interface artifacts.
Required Qualifications
Education
¢
Master's or PhD in Engineering, Science, Technology, or a closely related discipline.
Subject Matter Expertise
¢
Deep command of the first principles and analytical methods of your discipline, with
the ability to formulate graduate-level, multi-step problems and produce fully derived
solutions.
¢
¢
¢
Proficiency in both text-only problem formulation and the interpretation of technical
visuals, including diagrams, drawings, schematics, plots, and layouts.
Strong analytical skills, with fluency in mathematical derivation, dimensional analysis,
and numerical methods.
Excellent technical writing, with the ability to present reasoning in a structured,
exact, and reproducible manner.
¢
¢
Working proficiency in LaTeX for typesetting equations, symbols, and units.
Demonstrated background in academic or industrial research, university-level
teaching, or advanced problem authoring.
Preferred Qualifications (Bonus)
¢
¢
Familiarity with grading rubrics and Olympiad- or examination-style question design
Experience producing or working with industrial drawings, technical drawings,
schematics, or data visualizations.
¢
Peer-reviewed publications, patents, or professional certification

How to Apply
Submit your CV and indicate your area of specialization. Shortlisted candidates will complete a
two-part assessment: a text-only problem and a multimodal problem grounded in a technical
visual, each accompanied by a fully derived solution.

📌 SME (Noida)
🏢 Innodata India
📍 Noida