Data Science & Engineering Foundations

Context

My early roles took me through software testing, clinical statistics and insurance data. They taught me how to check software against detailed specifications, work with regulated data and turn analysis into something people could use.

Challenge

The details changed between roles, but the challenge was similar: turn product specifications, regulated clinical outputs or fast-changing insurance risk into reliable work.

Approach

Software Tester · TCP Lifesystems

June 2015 – August 2018 (Part-Time) · Bromley, Surrey

I worked three days a week during my A-levels, then returned during university holidays. The role was SQL-heavy: I took product specifications for new functionality or actuarial pricing changes and worked out how to test them thoroughly. I built Selenium tests for the user journeys and checked the results against the back-end data in SQL. Those tests became part of the regression suite, with SQL summaries included in the daily reports.

I also investigated production defects in SQL and reproduced them in test environments before checking the fix.

• Manual and automated software testing
• Bug tracking and defect management
• Test case development and execution
• Quality assurance process improvement

Clinical Statistician and Data Scientist · GlaxoSmithKline

September 2018 – August 2019 · Uxbridge, Middlesex

During my placement year at GSK, I split my time between the statistical data sciences and clinical research teams. I applied statistical methods to clinical trial data used in research and regulatory submissions.

Submission documents had to be regenerated whenever a dataset or output changed so that every version and timestamp remained current. Much of that work involved copying outputs into PowerPoint. I wrote an R tool that generated the submission decks from the underlying datasets, added the required timestamps and let people define their outputs once before rebuilding a deck in seconds.

• Statistical analysis of clinical trial data
• Experimental design and methodology development
• Regulatory documentation and reporting
• Collaboration with clinical research teams

Data Analyst · iPipeline

June 2020 – July 2021 · Bromley, Surrey

COVID-19 changed insurance risk quickly. Jobs that had previously been considered safe, such as supermarket work, could suddenly carry more risk, while the normal process for changing rates took three to six weeks. I worked with a client’s underwriting team to build a reporting database and Power BI dashboards. These gave underwriters a current view of risk and let them make rate changes in hours rather than weeks.

Underwriting rules classify each policy into a category. Referrals normally required two underwriters to agree on a rating before an offer could be sent. We built an XGBoost model that predicted a rating from the applicant’s answers. When its result matched the first underwriter’s decision, the quote could skip the second check and be sent immediately. This reduced drop-off while demand for life and income-protection insurance was high.

• Data analysis and reporting for insurance products
• Database design and SQL query development
• Dashboard creation and data visualisation
• Collaboration with business stakeholders

Results

Across the three roles, I learned to work carefully with specifications and regulated data, automate repetitive reporting and explain analytical results to the people making decisions.

Skills used

Analytics

• Data analysis and interpretation
• Statistical methods
• Reporting and visualisation
• Business intelligence

Statistics

• XGBoost model development
• Experimental design
• Hypothesis testing
• Clinical statistics

Software Quality

• Test planning and execution
• Regression analysis
• Quality assurance
• Bug tracking
• Process improvement