PSO Automated Scorer

Flagship

Automated evaluation, tiebreaker, and bracket ranking engine processing 4,000+ national Science Olympiad competitors across multiple elimination tiers.

RoleLead Scoring Architect
Platform
Server
Tech Stack

Problem & Constraints

Grading, applying complex tiebreaker matrices, and ranking 4,000+ high school student competitors across regional cluster eliminations within tight 2-hour event turnaround windows.

How It's Built

1. Automated Matrix Scoring Pipeline

Vectorized NumPy and Pandas matrix operations evaluating regional cluster answer keys, applying subject-weighted penalties, and computing tiebreakers in seconds.

Trade-off:Vectorized in-memory matrices replaced manual spreadsheet formula recalculations that previously hung for 45+ minutes.

dart
1
2
3
4
// Vectorized score computation with blueprint-weighted penalties
def compute_scores(raw_matrix: np.ndarray, answer_key: np.ndarray, weights: np.ndarray) -> np.ndarray:
    correct_mask = (raw_matrix == answer_key)
    return np.dot(correct_mask.astype(float), weights)

Hurdles & Solutions

Multi-Way Tiebreaker Deadlocks

Problem:Top national qualifiers frequently tied on total score, requiring recursive evaluation of difficulty-weighted question tiers and timestamp priority.

Resolution:Implemented a deterministic multi-key sorting algorithm evaluating total score, tier-3 problem counts, and verification check marks in sequence.

Results & Numbers

Processed scores and verified rankings for 4,000+ competitors with 100% accuracy and zero tabulation delays.

4,000+
Competitors scored and ranked
100%
Tabulation accuracy across elimination rounds
Retrospective:Building a web-based tabulation dashboard with live audit logs would make proctor cross-verification even faster.