Human Priority — National IT Championship
A national LLM competition category, run three years in a row.
We created and operate the official LLM category of the National IT Championship — curriculum, evaluation rubric and live operations for 10,000+ participants, built on industry standards.
Context
A national competition needed an AI category that measured real professional skill — not prompt tricks — and could survive scrutiny from universities, employers and sponsors.
The costly problem
No accepted benchmark existed for practical LLM competence at national scale; earlier attempts collapsed into either academic theory or marketing demos.
Baseline
Year zero: no category, no rubric, no task bank. Success defined as: industry-standard tasks, reproducible scoring, and participation at national scale.
Intervention
A task bank mapped to industry workflows (analysis, drafting, evaluation, tool use), a transparent rubric with human judging checkpoints, and a curriculum that lets schools and companies prepare.
Human responsibilities
- Human judges score all final-round work against the rubric; model-assisted pre-scoring never decides alone
- An appeals path lets participants challenge scoring — a few each year, all resolved on the record
- We train the judges annually; the rubric is versioned and published
Measured outcome
Limitations
Participation figures count registered participants as published by the organizer. The rubric measures competition performance; it is not a hiring certification.
A repeatable annual operation: task bank refresh, judge training, live events, appeals. Run with a small core team plus trained judges.
Next step
A follow-up category for agentic workflows is in design with the organizer.