Data Science & Analytics Consulting

Turning complex data into strategic insight

I build machine learning systems, NLP pipelines, and analytical tools that help organisations make better decisions. Specialising in financial analytics, risk assessment, and applied AI.

View Projects Get in Touch
14+
Years in Technology
Cambridge
Career Accelerator in Data Science, ML & AI (2025)
London
Based in the UK
About

Data science built on industry experience

I spent over 14 years in software testing at Sky, latterly as Test Manager, where I led cross-functional teams, optimised complex technical processes, and delivered a 25% reduction in resource demand through systematic quality engineering.

That experience taught me how large organisations actually work, where data gets lost, where decisions get made on instinct rather than evidence, and where machine learning can genuinely move the needle.

In 2025, I completed the Career Accelerator in Data Science with Machine Learning & AI (Certificate of Completion, University of Cambridge Institute of Continuing Education with FourthRev), formalising the analytical and engineering skills I now apply through RJ Data Voyage. My final employer project, a team prototype for a UK financial regulator, combined NLP sentiment scoring with financial-metrics extraction across public disclosures from systemically important banks.

I work with organisations that need practical, production-ready data solutions rather than slide decks and theory. If the problem involves unstructured text, complex classification, time-series forecasting, or turning messy data into clear strategic insight, I can help.

Industry-Tested Approach
14+ years in software testing and test management at Sky, delivering under real-world constraints with measurable outcomes.
Cambridge-Trained
Career Accelerator in Data Science with Machine Learning & AI, Certificate of Completion, University of Cambridge Institute of Continuing Education with FourthRev (2025). Grounding in ML, NLP and statistical methods.
Full-Stack Data Science
From data preparation and modelling to a documented, working prototype, such as a self-hosted AI system built for a client.
25%
Resource demand reduction at Sky
81
Public bank reports and transcripts analysed (G-SIB team project)
19
Published case studies
14+
Years in technology at Sky
Services

How I can help

Three focused ways I turn data and AI into working systems for organisations in regulated and data-heavy sectors - practical, validated, and built to run.

Data Science & ML
Predictive modelling, segmentation & forecasting
From a first model to production deployment - classification, clustering, forecasting and dashboards for data-heavy teams.
NLP & Text Analytics
Turn unstructured text into insight
Sentiment, topic modelling, entity extraction and summarisation across reports, transcripts and feedback.
Applied AI / RAG
Privacy-first local LLM & RAG
The capability of modern AI without sending your documents to a third-party AI provider - local models, RAG and query engines.
How I work

From first conversation to working system

Every engagement is fixed-scope and outcome-led - no open-ended billing, no black boxes.

01 - Discovery

Understand the decision

We start with the decision you need to make, the data you hold, and what a good outcome looks like.

02 - Scope

A fixed, defined proposal

Clear deliverables, timeline and success criteria agreed before any build begins.

03 - Build

Model, validate, iterate

Hands-on development against your real data, with progress you can see and question at each step.

04 - Handover

Production-ready

Documented code and a dashboard or interface your team can actually use - not a slide deck.

Featured Work

Selected projects

Client work and Cambridge coursework. Each case study states the method, the measured result and its limitations.

Financial Regulation / NLP

G-SIB Disclosure Analysis Prototype

81 reports and transcripts 17,292 text chunks scored

Supervisors monitoring systemically important banks rely heavily on long, unstructured disclosures and earnings-call transcripts. Reading them manually is slow and hard to compare across banks and quarters.

Team project for a UK financial regulator: I designed and built the core NLP pipeline scoring sentiment across three G-SIBs
FinBERT VADER GPT-2
View Case Study
Financial data analysis and risk assessment
Natural Language Processing

Customer Review Topic Modelling and Emotion Analysis

27,585 reviews analysed 2,031 anger-led reviews

A UK gym chain had thousands of free-text reviews across two public platforms and no systematic way to see what was driving negative feedback.

Topic modelling and emotion classification surfaced four main anger drivers: cleanliness, staff, equipment and fees
BERTopic Transformers LLM NLTK
View Case Study
Text analysis of customer reviews
Anomaly Detection / Operations

Ship Engine Anomaly Detection

422 readings flagged (2.16%) IQR rule

Engine sensor data arrives without fault labels, so abnormal readings have to be found without knowing in advance what a fault looks like.

An IQR rule across six sensors flagged 422 readings for review, compared against One-Class SVM and Isolation Forest
IQR Isolation Forest One-Class SVM PCA
View Case Study
Ship engine monitoring
Applied AI / RAG / Privacy

Private Local Documents AI

25+ file formats no third-party AI provider

Organisations handling confidential documents can't hand them to public AI services. They need the capability of a large language model on infrastructure they control.

Client prototype: a self-hosted LLM on a dedicated GPU server with a three-tier query engine and RAG semantic search, handed over with full documentation
Local LLM RAG Semantic Search GPU Python
View Case Study
Private local AI document processing on GPU infrastructure

Additional Projects

Further work spanning forecasting, deep learning, predictive analytics, and QA engineering.

Forecasting
Book Sales Demand Forecasting
Weekly demand for two long-selling titles forecast with SARIMA, XGBoost, LSTM and hybrids on a 32-week hold-out. A well-specified seasonal model was the strongest default; the hybrids did not beat it.
19.6% MAPE, SARIMA (Title B)
Predictive Analytics
Student Non-Completion Prediction
XGBoost at three stages of the student journey. Adding engagement data gave the best balance of accuracy and lead time; the final-stage result is flagged for a leakage check.
0.650 F1, engagement stage
Time Series
Crime Incident Forecasting with SARIMA
Daily incident data for a US city resampled, tested for stationarity and modelled with seasonal ARIMA, with a 52-week cycle confirmed and models compared on a hold-out year.
139.98 test MSE, best SARIMA
Financial Services
Bank Customer Churn Classification
Decision tree and random forest models predicting which retail bank customers will leave, with the trade-off between catching churners and raising false alarms made explicit.
0.61 churner F1, random forest
Statistical Analysis
Statistical Hypothesis Testing Framework
Six business questions tested with t-tests, ANOVA, chi-square and Mann-Whitney U, with normality checks driving test choice and three nulls honestly retained.
6 hypothesis tests at α = 0.05
Deep Learning
Neural Network Design and Hyperparameter Search
A Keras spam classifier built, tuned and evaluated over three exercises. Grid search over epochs and batch size did not beat the simple 64-32 baseline.
94.6% baseline test accuracy
QA Engineering / Test Automation
Automated PDF Compliance Test Harness
Layout and content checks for generated PDF letters, with per-file failure reasons and a colour-coded Excel report. Verdicts matched ground truth on a synthetic 300-letter corpus with five seeded defect types.
300 / 300 verdicts matched (synthetic corpus)
Business Analytics / ML
Customer Segmentation with Clustering
K-Means and hierarchical clustering of 68,300 retail customers, published as a learning exercise: silhouette scores were modest and one cluster held two-thirds of customers.
0.266 best silhouette score
Statistical Analysis / Insurance
Medical Insurance Cost Correlation Analysis
Pearson correlation analysis of what drives insurance charges across 1,000 anonymised records, set in a scenario where an employer weighs a BMI-linked benefits scheme.
r = 0.709 BMI correlation, strongest predictor identified
Marketing Analytics / ML
Decision Trees and SHAP: Bank Marketing
Tree-based models and SHAP on a public bank telemarketing dataset. Call duration dominated, but it is only known after the call, so the score describes outcomes rather than pre-call targeting.
0.9485 ROC AUC, Gradient Boosting (with leakage caveat)

Additional Work

Broader technical exploration outside my core financial services focus - methodology deep-dives and applied ML across other domains.

Natural Language Processing
Movie Review Sentiment Analysis
A classical baseline on binary sentiment, where Bag-of-Words and TF-IDF tied, then four recurrent networks on a harder five-class task, where the LSTM and GRU failed to learn.
38.5% five-class accuracy, Bidirectional LSTM
Business Analytics / Retail
Customer Loyalty Analysis
Regression and correlation analysis across 1,711 customers isolating which of quality, brand awareness and negative publicity actually drives loyalty.
r = 0.713 quality-loyalty correlation
Feature Engineering / Automotive
Automobile Price Prediction
Feature engineering, feature selection and PCA/t-SNE on a 205-vehicle dataset, comparing Random Forest and linear regression on a held-out test set.
R² 0.894 Random Forest, held-out test
Deep Learning Fundamentals
Neural Network Manual Backpropagation
One forward and backward pass of a small network written by hand in NumPy, without a framework, to show exactly what TensorFlow or PyTorch automates.
0 framework dependencies
Deep Learning / Model Evaluation
Optimiser Comparison: Adam vs RMSprop
Adam and RMSprop compared on the same small network for historical passenger-survival data, then regularisation, dropout and early stopping explored.
86.0% vs 83.2% test accuracy, RMSprop vs Adam
Expertise

Core competencies

Six areas where I deliver the most value, each backed by project work and production experience.

Machine Learning

Supervised and unsupervised methods including XGBoost, Random Forests, SVM, ensemble techniques, and clustering (K-Means, hierarchical). End-to-end pipeline development.

Natural Language Processing

Transformer-based models (FinBERT, BERT, GPT-2), topic modelling (BERTopic), sentiment analysis, text classification, and domain-specific NLP for financial and customer data.

Forecasting and Time Series

ARIMA/SARIMA, XGBoost, LSTM networks, hybrid models and decomposition techniques. Applied to demand planning and incident forecasting.

Data Visualisation and BI

Interactive dashboards and visual analytics using Matplotlib, Seaborn, Plotly, and Power BI. Clear, decision-ready presentation of complex findings.

Statistical Analysis

Hypothesis testing (parametric and non-parametric), correlation analysis, regression, and model validation.

Anomaly Detection

Isolation Forest, One-Class SVM and statistical rules (IQR) for flagging irregular readings in operational sensor data, with results prioritised for expert review.

"Raquel consistently demonstrated outstanding leadership throughout our final project in the Cambridge University Data Science Program. She distinguished herself by proposing innovative solutions, guiding the team toward consensus, and ensuring decisions were implemented effectively. She combines strong technical expertise with sound judgment, and excels at aligning teams around clear objectives."

S.S. - Business Information Analyst • Cambridge Data Science Career Accelerator cohort
Insights

Notes from the work

Practical notes on applied AI and data science, drawn from real project work.

Applied AI / RAG
Running a private RAG system on confidential documents
How to give your team AI over sensitive documents, with answers grounded in your own files and nothing leaving infrastructure you control.
Contact

Let's work together

I partner with organisations on data science, analytics engineering, and applied ML projects - from a single well-scoped engagement to an ongoing retainer. If you have a complex data challenge, get in touch to discuss how I can help.

Based in London, available for remote and hybrid work across the UK.

GitHub
LinkedIn
London, United Kingdom