About Me
Bio
Data & Automation Analyst at Ditto Music, where I've built fraud detection pipelines, automated QC systems, and royalty reporting infrastructure from the ground up — saving hundreds of hours of manual work in the process. I'm drawn to problems where data can genuinely change an outcome, whether that's catching fraudulent streams or predicting what a customer does next. Currently developing my machine learning toolkit with a focus on predictive modelling and decision science.
Current Focus
- RoleData & Automation Analyst
- SpecialtyAnalytics & Modelling
- LocationLiverpool, UK
- EducationUOC — Chester
Experience
Data & Automation Analyst
Ditto MusicDitto Music's first dedicated analyst, building fraud detection, automated QC, and commercial reporting infrastructure from scratch across a catalogue of 2M+ artists.
- Built DEEPFRAUD, a 6-signal fraud detection pipeline flagging ~3,000 suspicious accounts and 40M+ streams per month.
- Drove AutoQC automated pass rate from 1% to 19%, processing 20,000+ releases per week.
- Designed a weekly top-1,000 earner tracking pipeline across a 2M+ artist database, used by global RLS clients.
- Built multi-source Python ETL pipelines pulling from Elasticsearch, MySQL, and internal databases for rights and royalty reporting.
- Developed a confidence scoring system combining multiple fraud signals into a single ranked output for the fraud review team.
BSc Computer Science
University of ChesterStudied core computer science fundamentals with a focus on data, machine learning, and software development — culminating in a dissertation on fraud detection using supervised and unsupervised ML.
- Built and evaluated supervised (Random Forest) and unsupervised (Isolation Forest) models for fraud classification as part of dissertation research.
- Developed full-stack web applications using Python, Flask, HTML and CSS.
- Applied NLP techniques including transformer models (T5) and entity extraction with spaCy.
- Gained foundational experience in data preprocessing, feature engineering, and model evaluation metrics.
- Worked with SQL for data querying and manipulation across multiple projects.
- Introduced to ETL concepts through real-time data pipeline and dashboard projects.
Projects
DEEPFRAUD
Fraud detection pipeline analysing 40M+ streams/month across Ditto Music's catalogue. Six weighted signals produce a unified artist risk score.
AutoQC Reporting Suite
QC analytics pipeline across 20K+ weekly releases at Ditto Music. Data-driven iterations drove the automated pass rate from 1% to 19%.
Account Performance Dashboard
Weekly pipeline surfacing Ditto Music's top 1,000 royalty earners from 2M+ artists. Powers commercial decisions for global RLS clients.
Fraud Detection Tool
Dissertation ML app detecting fraud in CSV transaction data. Auto-selects Random Forest or Isolation Forest based on whether labels are present.
Loan Default Prediction
ML web app predicting loan default risk from applicant financial data. Returns a confidence score and the key risk factors driving the decision.
Text2SQLAI
NLP tool converting plain English questions into SQL using a fine-tuned T5 transformer. Rule-based fallback ensures reliability when the model underperforms.