AI Code Quality Analyzer
Machine Learning & AST-Based Static Analysis Engine
Automated code review tools often flag basic stylistic conventions while missing deeper structural complexity, nested cognitive load, and maintainability risks.
I built a dual-layer analysis engine that parses Python source into Abstract Syntax Trees in memory, extracts cyclomatic complexity and Halstead metric vectors, and uses a trained scikit-learn classifier to predict code maintainability.
IN-MEMORY AST INSPECTOR
sub-50ms parse
def evaluate_maintainability(source_ast):
tree = ast.parse(source_ast)
metrics = compute_cyclomatic_halstead(tree)
return classifier.predict(metrics.vector)
CYCLOMATIC
1.0 (Clean)
HALSTEAD
142.6 E
PREDICTION
High Quality
PHASE 01
In-Memory AST Parsing
Accepts Python source via web UI and parses syntax trees in memory without disk write overhead.
PHASE 02
Metric Computation
Traverses nodes to calculate McCabe cyclomatic complexity, Halstead volume, and nesting depth.
PHASE 03
ML Classifier Inference
Feeds normalized metric vectors to a trained model to predict maintainability class.
