Full-Stack · AI
AgriCrop — AI-Powered Agricultural Platform
"Integrating AI with web technologies to automate agricultural intelligence at scale."
A production-grade full-stack web platform that combines AI with modern web technologies
to automate tasks, provide intelligent responses, and manage data securely.
Built with a scalable architecture that separates concerns across presentation,
business logic, and data layers. Demonstrates end-to-end system design from
database schema to intelligent API endpoints.
Tech Stack
Python
AI/ML
Web APIs
SQL
Full-Stack
Authentication
Machine Learning · Fintech
Intelligent Credit Card Fraud Detection
"Detecting financial fraud on highly imbalanced datasets using ML classification pipelines."
An end-to-end ML system that identifies fraudulent credit card transactions on
real-world highly imbalanced financial datasets. Key engineering challenges include
handling extreme class imbalance, selecting appropriate evaluation metrics
(Precision-Recall over Accuracy), and applying data mining techniques to
extract meaningful signal from noisy transactional data.
Tech Stack
Python
Scikit-learn
Pandas
NumPy
Imbalanced-learn
Matplotlib
Data Engineering · Analytics
Revenue Intelligence Platform
"Modular business analytics engine with ARIMA forecasting and RFM customer segmentation."
A modular, production-structured Python platform that processes transactional data,
performs RFM-based customer segmentation to identify high-value customer cohorts,
analyzes product performance across dimensions, and forecasts future revenue using
ARIMA time-series models. Designed with clean module separation and reusable
pipeline components.
Tech Stack
Python
ARIMA
Pandas
NumPy
Matplotlib
Statsmodels
ML Fundamentals
Logistic Regression From Scratch
"Implementing gradient descent and sigmoid optimization in pure NumPy — no black boxes."
Built a complete Logistic Regression classifier from first principles using only NumPy —
implementing the sigmoid function, gradient descent optimization, binary cross-entropy loss,
and evaluation metrics (accuracy, precision, recall, F1). Performance validated against
Scikit-learn's implementation on the same dataset. Demonstrates deep understanding of
the mathematics behind modern ML algorithms.
Tech Stack
Python
NumPy
Matplotlib
Gradient Descent
Scikit-learn
Backend · Systems
Smart Task Scheduler — Java Concurrency Engine
"Priority-based concurrent task scheduling with file persistence and clean layered architecture."
A production-quality task scheduling engine built in Core Java using custom Comparators
for priority ordering, Java multithreading for concurrent task execution, and file-based
persistence for durability across restarts. Implemented with a clean layered architecture
separating domain logic from infrastructure concerns — demonstrating systems engineering
maturity beyond typical academic projects.
Tech Stack
Java
Multithreading
Custom Comparator
OOP
File I/O
Layered Architecture
Data Engineering · BI
Smart Logistics Supply Chain Analytics
"End-to-end logistics analytics with SQL querying, Python pipelines, and Power BI dashboards."
A comprehensive data analytics project analyzing supply chain performance metrics
across a logistics operation. Covers the full data workflow: raw data ingestion,
Python-based cleaning and preprocessing, SQL analytical querying for business
insights, and an interactive Power BI dashboard for stakeholder reporting.
Demonstrates practical data engineering skills that translate directly to
enterprise analytics roles.
Tech Stack
Python
SQL
Power BI
Pandas
Data Visualization
EDA