The work that best shows how I combine finance, data engineering and analytics.
An end to end Python pipeline that loads NHS England financial data into a MySQL star schema, with analytical views, key performance indicators and Power BI reporting across 206 trusts.
Explores sales, discounts and profitability by category, segment and region, then trains a logistic regression model to separate profitable orders from loss making ones.
Tests whether changes in NHS waiting lists come from treatment capacity or from the number of patients starting treatment pathways, using NHS England Referral to Treatment statistics.
Dashboards and data models that turn public and financial data into decision-ready views.
Rebuilds the visa, asylum, settlement and citizenship journey as a Power BI data model, using Home Office statistics from the September 2025 release alongside Office for National Statistics economy data.
Interactive report on the QS World University Rankings 2025 to 2026, comparing institutions and countries on the published ranking measures.
Python workflows for financial data, applied machine learning, model deployment and automated reporting.
A modular data engineering project that gathers economic indicators and market prices from public sources into reproducible research datasets.
Seven numbered modules that turn finance and business questions into prediction tasks, from regression and classification to decision trees and ensembles, finishing with a deployed model.
MySQL data models, analytical SQL queries and Python integration.
Three modules from MySQL basics to stored programs on the ClassicModels sample database, run from Python so every query returns a clean pandas data frame.
Collects public comments on two UK immigration videos through the YouTube Data API and stores them in MySQL, as the groundwork for sentiment analysis.
Prediction, classification and explainability work on business data.
Investigates how product features drive price, using a first case study of 53,940 diamonds, with a Streamlit app that estimates a price from quality grades and measurements.
Explores how job title, experience, location and remote work relate to salary, comparing a PyTorch neural network with linear regression, random forest and XGBoost.
Frames lead conversion as a classification problem on the Kaggle leads dataset, with a plan for an interpretable logistic regression baseline and lead scoring bands.
Profiles mobile shop customers and trains an XGBoost classifier to tell repeat phone buyers from first time customers, using demographics, price and Facebook engagement.
Retrieval, agents and automation built on large language models.
A retrieval agent built step by step, from keyword search over FAQ data to a reusable retrieval augmented generation pipeline and an agent loop with function calling.
Streamlit dashboard that retrieves headlines and articles from NewsAPI, with category and keyword search and interactive charts.
Flask web app that reads an uploaded PDF and answers questions about it, using LangChain, OpenAI and a FAISS vector index for semantic search.
Gmail agent that fetches messages, sorts them into categories with a rule based classifier and drafts replies, using OpenAI or a built in fallback.
Docker Compose set up for an agent that talks to a locally running Ollama server, so language models run on your own machine.
Get In Touch
Interested in discussing Finance Analytics, Data Engineering, Power BI implementation, or Python-based financial solutions? I'm based in the UK and available for immediate engagement.
Start a Conversation