Shahid Ul Islam | Portfolio

TrustLens

Open-source ML reliability framework for auditing models beyond accuracy.

“Your model has 92% accuracy. That may still be unsafe.”
Evaluating calibration, fairness, failure patterns, and deployment risk in one unified pipeline.

Composite Trust Score 88/100

~10.3k Lines of Code

16 Test Suites

25 Source Files

TrustLens is an open-source ML reliability and diagnostics framework for evaluating model trustworthiness beyond accuracy. It analyzes calibration, fairness, failure patterns, representation quality, and deployment risk using a unified API and composite Trust Score system.

Open Source PyPI Package CI/CD Plugin Architecture ML Reliability Production Ready

Python 3.9–3.13

Multi-module architecture

Extensible Plugin System

Automated Benchmarking

Modular Architecture

Public API Layer

Core Engine

Metrics Modules

Visualization

Plugin Registry

TrustReport

Trust Score Engine

Diagnostics

trustlens_audit.py

from trustlens import analyze
 
report = analyze(
    model,
    X_test,
    y_test,
    y_prob=model.predict_proba(X_test)
)
 
report.show()

              TRUST SCORE: 88/100 [B] 

              Assessment : Good Trust - minor issues to address 

              Score Summary: 

              Base Score        : 92 

              Penalties Applied : -4.0 [Calibration (-4.0)] 

              Final Score       : 88

Open Source Excellence

Built as a collaborative open-source ML infrastructure project.

Explore TrustLens View on GitHub View on PyPI Watch Demo

Projects &

Practical Work

Applying machine learning, data science, and analytics to real-world problems — from healthcare AI to business intelligence dashboards.

Respiratory Disease
Classification

Advanced transfer learning with explainable AI (XAI) to classify respiratory conditions from medical imaging with clinical-grade precision.

Case Study

Breast Cancer
Prediction

Predictive analytics framework leveraging statistical classification techniques to assist in early-stage breast cancer diagnosis.

Case Study

CardioSense
AI

AI powerd cardiovascular decision support system that not only predicts heart disease risk but also explains the reasoning behind predictions

Case Study

Apple Sales
Dashboard

Comprehensive data visualization ecosystem providing real-time market insights and product performance analytics for Apple ecosystem sales.

Case Study

UP NEXT

Machine Learning

From First Principle

We don't just see the data; We understand the mathematical space where it exists.

Gradient Descent

Optimization from first principles — gradients, learning rates, convergence, and modern optimizers.

Read the Article

Lagrange Multipliers

From geometry to least squares — understanding constrained optimization mathematically.

Read the Article

Bias Variance TradeOff

The fundamental trade-off between model simplicity and prediction accuracy.

Read the Article

Logistic Regression

Classification through probability, sigmoid functions, and decision boundaries.

Read the Article

Explore All Topics

Architecting Digital Impact

TrustLens