Projects
Below are some projects which I have been working on.
Automated Weather ETL Pipeline
A full-stack data engineering project that automates the extraction, transformation, and loading (ETL) of meteorological data. Utilising Python and the Open-Meteo API, the pipeline runs via GitHub Actions on a 6-hour cron schedule, injecting structured JSON data into a static Eleventy environment to power a real-time weather dashboard.
- End-to-End ETL Lifecycle: Completed pipeline covering Extraction (API to Python), Transformation (data cleaning and date formatting), and Loading (JSON injection into a static site).
- Automation: Using GitHub Actions for Automation to retrieve informations
Achievement Prediction Ensemble Models Project
This project investigates the key drivers of reading and mathematics performance among kindergarten students. Using a multi-level dataset (student, teacher, and school levels), we employed a variety of econometric and machine learning models to identify the most accurate predictors of academic success.
- Top Model: XGBoost outperformed all other models with a combined Mean Squared Error (MSE) of 990.22.
- Primary Determinant: Socioeconomic Status (SES), proxied by eligibility for free/reduced-price lunch, was the strongest predictor of academic outcomes.
- Environment Matters: Students in rural areas showed higher proficiency in reading, while those in suburban areas excelled in mathematics.
- Class Size: Consistent with Project STAR findings, smaller class sizes had a notable positive impact on performance.
IBM Machine Learning Project
Analysing datasets and solving real-world problems in collaboration with IBM, utilizing a suite of enterprise tools to deliver solutions.
- Data Processing: Using Python and Pandas for cleaning and transforming data.
- Predictive Modelling: Building and validating models with visualisations.
- IBM Watsonx & LLMs: Exploring AI platforms and Large Language Model applications.
- RAG Optimization: Implementing Retrieval-Augmented Generation techniques.
Automated Geographic Data Extraction
Developed an automated solution to extract, process, and analyse geospatial data from a RESTful API and export it into a structured CSV format.
- Key features: API integration, multi-format handling (JSON/XML/CSV), and error handling for rate limits.
- Challenges: Handling unstructured data and improving scraping efficiency while staying compliant.
Personal Fitness Dashboard Project
Built an interactive Power BI dashboard to track and analyse personal fitness data (steps, calories, distance) for actionable insights.
- Designed a Power BI dashboard tracking health-app metrics.
- Created interactive visuals showing monthly and yearly trends.
- Optimised the data pipeline to improve clarity and usability.
My Resume Website
Personal resume website designed to showcase skills and projects with a responsive UI.
- Built using HTML, CSS, and JavaScript.
- Responsive layout with interactive elements.
- Smooth navigation and clean UI design.
SQL & Python Pokemon Managing System
Built Python scripts like a personalised Pokédex to extract and display information from a Pokémon database.
- Python scripts: Detailed Pokémon information extraction.
- SQL & PLpgSQL: Enhanced retrieval and data management functions.
Huffman Tree Encoder in C
Created a Huffman Tree Encoder using binary trees in C for variable-length encoding.
- Binary Search Tree (BST): Efficient data storage.
- Huffman Tree: Encoding based on character frequencies.