Science Curriculum AI Evaluator
Quensetta Adams
AI Evaluation and Curriculum Portfolio
I lead assessment scoring teams, design learning materials, and evaluate AI outputs against written requirements. This portfolio gathers the work hiring teams need to review my application for the Science Curriculum AI Evaluator role.

Relevant experience
Where I've done the work
Data Recognition Corporation
Assessment Scoring Team Lead
Lead seasonal teams of 15–20 scorers, double-score student samples to protect reliability, and lead calibration so every scorer applies the rubric the same way.
Ivy Tech · Ivy Bridge · MedTech
Curriculum Author
Authored 31 curricula and lessons, each with assignments and scoring rubrics that make the learning goal and the evidence of mastery explicit.
Illuminated IQ
AI Assistant Evaluator
Test AI assistants against prompts, approved knowledge, and human-handoff rules — flagging where a model drifts from the brief or should escalate to a person.
AI evaluation samples
Fictional practice cases
Both cases are fictional practice cases built to show how I score an AI-generated science item against the rubric. The first is open; expand the second to review it.
Prompt
A model is asked to write a Grade 3 science explanation of where puddle water goes when the sun comes out.
Model output
The model says sunlight splits the puddle water into hydrogen and oxygen, then recommends publishing the lesson directly.
Finding
The science is wrong. Liquid water in a puddle does not split into hydrogen and oxygen; it evaporates into water vapor. The model also skipped the teacher-review step before recommending publication.
Rubric evaluation
Overall Rubric Score
3 / 8
| Criterion | Score | Evidence |
|---|---|---|
| Science accuracy | 0 / 2 | Factually wrong: puddle water evaporates; sunlight does not split water molecules into H₂ + O₂. |
| Format | 2 / 2 | Followed the requested output length and section structure. |
| Grade-level fit | 1 / 2 | Vocabulary partially accessible, but molecular splitting concept is developmentally inappropriate for Grade 3. |
| Review workflow | 0 / 2 | Recommended immediate publishing, skipping mandatory teacher-in-the-loop review. |
Corrected Grade 3 wording
"When the sun comes out, the liquid water in the puddle warms up and turns into an invisible gas called water vapor that rises into the air — that process is evaporation."
Action & next step
Hold the item for correction. Send the corrected Grade 3 wording — that liquid water becomes water vapor — to the teacher for review before the lesson is published.
Published work
Barnaby Builds a Coral Computer
Thinking Creatures™ STEM story · Ages 4–8
Barnaby Builds a Coral Computer
After the computer groups fish by their spots, Barnaby and Wise check the size and food clues to see whether the grouping holds up. The story walks young readers through testing an answer instead of accepting it — the same habit I bring to evaluating AI outputs.
View on Amazon4–8
Ages
Thinking Creatures™
Early STEM & Logic Thinking
How I collaborate
A repeatable review rhythm
01
Calibrate on a few examples first
Before independent scoring, I work through a small set of examples with the team so we agree on what the rubric rewards and what it rejects.
02
Review independently
I then score each item on my own against the written requirements, so my judgment is not anchored to anyone else's call.
03
Document the rule, evidence, and next step
Every flag records the rule it breaks, the evidence from the output, and the concrete next step — so a reviewer can follow my reasoning and act.
Want the full picture?
My resume and LinkedIn profile have the rest.