Projects | Mostafa ElAnany

ChessLens - Chess Analytics Platform

End-to-End Chess Analytics with dbt, DuckDB & Stockfish

Chess.com gives you a rating graph and a win rate. It doesn't tell you when you play your worst chess, whether losing streaks actually make you play worse, or how many games in a row is too many. ChessLens answers those questions. Any chess.com user can enter their username and get their full analysis in about a minute.

Project Type

Data Engineering / Analytics

Data Model

8 dbt models, 18 schema tests, medallion architecture

Tested Across

9,000+ games, multi-user

Key Features

dbt medallion architecture (bronze → silver → gold) with incremental models and 18 schema tests
Tilt Tracker: win rate analysis by consecutive losses to detect tilt patterns
When To Play: heatmap of win rate by day of week and time of day
Opening Lab: monthly opening win rate trends with improving/declining/stable detection
Know Your Opponent: win rate segmented by opponent rating buckets, upset win tracking
Session Insights: session detection (30-min gap) with rating change by session length
Game Replay: move-by-move board with eval bar, Stockfish analysis at depth 20, move classification dots
Dagster daily orchestration, full pipeline runtime under 30 seconds
Python ingestion supporting historical backfill and incremental refresh

Python SQL dbt-core DuckDB Dagster Streamlit Plotly Stockfish python-chess Medallion Architecture Incremental Models

Live Demo View Repository Read Article

Formula 1 Real-time Data Engineering Pipeline

Modern Data Architecture for Racing Analytics

This project involves building a data pipeline that processes Formula 1 racing data using modern data engineering tools. While the source data is batch, I implemented real-time architecture for learning purposes. The system integrates Python for data collection, PostgreSQL for storage, Kafka and Debezium for streaming, and DuckDB with dbt for analytics.

Project Type

Data Engineering / Streaming

Skills Applied

Stream Processing, CDC, Data Modeling

Learning Focus

Event-Driven Architecture

Key Features

Complex data collection from multiple F1 data sources
OLTP PostgreSQL database with OLAP DuckDB for analytics
Real-time data streaming with Kafka and Debezium
Change Data Capture (CDC) for event tracking
Slowly Changing Dimensions (SCD) Type 2 implementation with dbt
Medallion architecture for data quality management
Asynchronous Python programming for efficient API calls

Python PostgreSQL Kafka Debezium DuckDB dbt Airflow Docker Grafana Prometheus Poetry Tableau Data Modeling

Read Article Repository Coming Soon

Driver Attention Detection

Deep Learning for Driver Safety

This project focuses on using computer vision and deep learning to determine whether a driver is driving safely with proper attention. I implemented and compared 7 different deep learning models to find the most accurate approach for this critical safety application.

Project Type

Computer Vision / Deep Learning

Skills Applied

Deep Learning, Model Comparison, Web Deployment

Learning Focus

Neural Network Architectures

Key Features

Implemented 7 different deep learning architectures
Comparative analysis of model performance and accuracy
Built an interactive Streamlit web application
Real-time image classification for driver safety assessment
User-friendly interface for uploading and analyzing images

Python Deep Learning Computer Vision Streamlit TensorFlow Image Classification

Live Demo View Repository

Cats and Dogs Classifier

Image Classification with CNN and Transfer Learning

This project demonstrates different approaches to image classification by comparing custom Convolutional Neural Networks (CNN) with pre-trained models. The goal was to accurately classify images as either cats or dogs and make this capability accessible through a user-friendly web application.

Project Type

Deep Learning / Image Classification

Skills Applied

TensorFlow, Transfer Learning, Model Deployment

Learning Focus

Custom vs Pre-trained CNN Models

Key Features

Custom CNN implementation from scratch
MobileNet pre-trained model implementation with transfer learning
Achieved 97% accuracy with the pre-trained model
Streamlit web application for public use
User-friendly interface for image upload and classification

Python TensorFlow CNN MobileNet Transfer Learning Streamlit

Live Demo View Repository

H&M Personalized Fashion Recommendations

Exploratory Data Analysis for Retail

This project involves a detailed exploratory data analysis (EDA) on H&M dataset to understand customer preferences and shopping patterns. The insights derived from this analysis can be used to create personalized fashion recommendations, enhancing customer experience and driving sales.

Project Type

Data Analysis / EDA

Skills Applied

Pandas, Data Visualization, Statistical Analysis

Learning Focus

Retail Data Pattern Recognition

Key Features

Comprehensive exploratory data analysis
Customer segmentation based on purchasing patterns
Product category analysis and trends
Statistical analysis of customer behavior
Visualization of key insights and patterns

Python Pandas EDA Data Visualization Statistical Analysis Retail Analytics

View Repository

End-to-End Data Engineering Pipeline

Modern Data Stack Implementation with ELT Approach

This project showcases my expertise in data engineering by implementing a complete data pipeline using modern technologies. The primary objective was to generate synthetic data and integrate it with various tools to demonstrate an automated ELT approach with DevOps practices.

Project Type

Data Engineering

Skills Applied

ETL, Cloud Infrastructure, Orchestration

Learning Focus

Modern Data Stack

Key Features

Synthetic data generation using Python scripts
Implemented multiple CTEs for the data mart
Created a Snowflake schema for effective organization
Automated the entire process with Airflow
Utilized AWS services (EC2, S3, Lambda, SNS, Secret Manager)
Deployed Docker and Docker-Compose for container orchestration
Used dbt Core for data transformation and modeling

AWS Docker Airflow Snowflake dbt Python Pandas ELT

View Repository

My Projects

ChessLens — Chess Analytics Platform

Formula 1 Real-time Data Engineering Pipeline

End-to-End Data Engineering Pipeline

Driver Attention Detection

Cats and Dogs Classifier

H&M Personalized Fashion Recommendations