Driven by curiosity and a love for problem-solving & lifelong learning.
I specialise in leveraging ML & rigorous statistical analysis to uncover hidden patterns, drive data-informed decisions and build practical AI solutions. If you have a complex challenge that you'd like to tackle and are looking for a collaborative and reflective contributor - I'm your guy.
My most recent experience has been as a Data Analyst at Dayforce and as a Data Scientist for Catholic Crosscultural Services & The BRIDGE. In both roles, I have translated complex data into actionable insights, targeted business improvements and built practical data-driven AI solutions.
Core competencies & libraries I've consistently worked with:
Python
R
PostgreSQL
Power BI
Excel
PyTorch
Scikit-learn
SciPy
Matplotlib
Forecast
Built an operational personal AI assistant that connects conversational planning with real execution across time. David keeps context grounded in goals, weekly state and decision logs, while injecting live Google Calendar context to make practical scheduling suggestions. Uses explicit approval flows before any calendar write, supports session-aware memory, and automates recurring check-ins to reduce decision fatigue and keep priorities aligned week over week.
A customisable system for tracking ML models' performance metrics, detecting data drift, while supporting hot-reloading model capabilities and low-latency FastAPI inference. Designed to ensure production ML models remain reliable and deliver consistent predictions over time. Features include: alerting via Discord webhooks, dashboard visualisation, streamlined and simple deployment.
Identified VARX as the most accurate model for cocoa futures price prediction through a comparative study of time‑series forecasting methods (e.g. ARIMA-GARCH & ETS). Enabled scalable testing and deployment through scripting model fitting, back-testing and evaluation using the forecast, rugarch, and vars packages
Explored the use of masked autoencoders with teacher-student training in visual classification tasks. Created a synthesis on Transformers, Autoencoders, Gaussian Processes & Skip-gram models.
Built an operational personal AI assistant that connects conversational planning with real execution across time. David keeps context grounded in goals, weekly state and decision logs, while injecting live Google Calendar context to make practical scheduling suggestions. Uses explicit approval flows before any calendar write, supports session-aware memory, and automates recurring check-ins to reduce decision fatigue and keep priorities aligned week over week.
I'm always open to discussing new projects, creative ideas, or opportunities to be part of your visions. Any connections are always appreciated - feel free to reach out!
