journal
Notes on backend engineering, fintech, API design, and applied AI. Long-form pieces — published when there's something worth saying.
Featured
-
Hybrid Deep Learning for Solar Energy Holding Capacity in Bangladesh
A stacked ensemble combining Random Forest, XGBoost, Gradient Boosting, AdaBoost and a Neural Network meta-learner reaches ~95% accuracy on Bangladesh solar-irradiance data.
-
Sentiment Analysis of COVID-19 Post-Vaccination Discourse in Bangladesh
Classical ML, ensemble classification and LSTM with Word2Vec embeddings applied to Bengali / English-mixed social-media chatter — best 78.8% accuracy.
All posts
17 posts-
Building a Dispute Automation System in .NET 9
A production-grade .NET 9 Web API for utility-bill disputes — layered architecture, repository + service-invoker patterns, two-level logging, and a step-by-step partner onboarding flow.
-
How Banks and FinTechs Integrate with Payment APIs
The full integration lifecycle — partnership onboarding, IP whitelisting, payer-type classification, routing, reconciliation, and OTC agent networks — from inside the payments aisle of a tier-1 bank.
-
How I Built Payment APIs: Architecture, Authentication & Integration
Notes on production-grade payment REST APIs — token authentication, two-phase commit, bill locking, idempotency keys, and the dispute-resolution patterns that actually hold up in audit.
-
Analyzing E-Commerce Sales Trends with Time Series Forecasting
A deep dive into seasonal patterns and predictive modeling for online retail sales using Python and statistical methods including ARIMA.
-
Machine Learning Model Comparison: Finding the Best Algorithm
Comprehensive evaluation of five ML algorithms on a customer-churn classification problem — comparing accuracy, precision, recall, and computational efficiency.
-
Data Pipeline Architecture: Building Scalable ETL Systems
Designing and implementing a robust data pipeline using Apache Airflow — handling millions of records daily with error handling, retries, and monitoring.
-
Deep Learning for Image Classification: A Practical Guide
Building and training a convolutional neural network using TensorFlow and Keras to classify images with 95 %+ accuracy on CIFAR-10.
-
SQL Optimization: Improving Query Performance by 10×
Analyzing a 45-second reporting query in production and implementing indexing, materialized views, and query rewriting to cut execution time to 4.2 seconds.