• Trained multiple machine learning models on historical stock price data to predict 1, 2 and 4-week future prices.
• Compared the performance and accuracy of a Long Short Term Memory neural network with a traditional multiple linear regression model, achieving an R2 error of 99.5% and 98.9% for the respective models.
• Implemented in Python using pandas, keras and sk-learn.
GitHub
• Developed a social media monitoring engine on Azure for a client from Microsoft as part of an Agile team.
• Implemented user stories, regular sprints, retrospectives, issue tracking, and a wiki.
• Configured the engine to monitor Twitter, Instagram, Reddit and GitHub accounts for specific post content, location and date/time.
• Built through a C# .Net web app, complete with automatic emails and HTTP requests via REST APIs.
• Created a VBA macro to compute a business client's investment interest according to asset value cash flows, required IRR, and simulation date inputs.
• Improved precision of IRR to 99.9% and automated a previously manual calculation process.
• Debugged and improved unit/integration tests in Python as part of a fully automated software testing suite for printer drivers, reducing overall test suite runtime by 30%.
• Utilised CI/CD pipelines in Jenkins to concurrently manage 24 different driver testing suites.
• Remodelled 500+ automated test descriptions and metadata to follow consistent design standards.
• Created over 100 new Python & JavaScript tests on virtual Linux machines using Mercurial.
• Optimised SQL statements to improve query performance by 20% and created a relational database search interface using Microsoft Access.
• Coded patient register form in HTML & CSS to more accurately record patient data.
• Transformed company data into graphs and diagrams using Microsoft PowerBI.
• Worked as part of a team to unify multiple duplicated user databases into a single table schema.
• Efficiently merged and updated over 10,000 user profiles in Salesforce.
