Assessment of Pakistan's Public-Health Data Infrastructure
Azra Analytics performed a countrywide evaluation of Pakistan's public-health data systems with U.S. CDC backing. The initiative encompassed technical review of the National Health Data Center, capacity surveys across 124 districts, and fieldwork at 81 health facilities spanning 39 districts. The team examined data flows, reporting approaches, infrastructure conditions, data quality, duplication concerns, and disease-surveillance mechanisms, then offered strategic recommendations for system enhancement and integration.
Expense Anomaly Detection
Azra Analytics developed an anomaly-detection framework for five years of expense transactions from a major leasing organization. The system combines rule-based controls with Isolation Forest, Histogram-Based Outlier Scores, and denoising autoencoder models to identify duplicate patterns, suspicious amounts, structural rarity, and unusual combinations of transaction features. The framework includes data preparation, model evaluation, multi-model anomaly comparison, explainability outputs, stakeholder reporting, and deployment planning for both batch and individual transaction reviews.
Integrated Data and Safety Platform for Transportation
Azra Analytics designed a technical architecture for a trucking and logistics company that integrates operational data from multiple Google Sheets into a centralized data warehouse, automated ETL pipeline, REST API, and dashboard-ready analytics layer. The project also uses millions of federal inspection records to model weigh-station activity, predict inspection outcomes and depth, and identify likely violation categories — giving managers, dispatchers, and drivers more timely, consistent, and actionable information.
Data Ingestion Pipeline for PDFs and Document Images
Azra Analytics designed and prototyped an on-premises AI platform that converts digital and scanned bank statements into validated transaction data and automated financial reports, using document classification, OCR, large language models, deterministic parsers, arithmetic validation, confidence scoring, and human exception review — reducing manual data entry, strengthening auditability, protecting sensitive financial data, and establishing a structured foundation for future risk analytics and decision support.
