Translating data into decisions that move enterprise.
Data analyst with 3+ years across SaaS automation, consulting and analytics. I combine a
technical foundation in SQL, Python, R and Tableau with a consulting track record at
HighRadius and British Telecom — turning complex datasets into commercially useful
recommendations for stakeholders, not just dashboards.
3+ YrsData Analyst & Consultant
MScBusiness Analytics · Warwick
03 — Skills
Functional, technical, and methodological.
Functional Competencies (8)
Data AnalysisBusiness Process MappingRequirements GatheringERP ImplementationUAT & WorkshopsSegmentationStakeholder ManagementBusiness Case Development
Data (9)
Machine LearningCRISP-DMClustering (K-Means, GMM, Hierarchical)Dimensionality Reduction (Factor Analysis)Optimization / Integer ProgrammingStatistical TestingPredictive Modelling (XGBoost, LightGBM, SVM, Random Forest, Deep Learning)Feature EngineeringRAG stack (FAISS, BM25, RRF, Cross-Encoder, HyDE)
Tools (8)
SQLPythonRTableauExcelSAPOraclePyomo
05 — Projects & research
Selected work outside the day job.
Needs-Based B2B Customer Segmentation for BT Business
MSc Dissertation
RCRISP-DMK-MeansFactor AnalysisGMMSegmentation
Led an end-to-end B2B segmentation study on a 1,102-respondent BT Business survey to improve upon blunt firmographic targeting. Reduced 80 behavioral items to 8 orthogonal factors via exploratory block-wise factor analysis. Evaluated Hierarchical, GMM, and K-Means clustering across multiple segment solutions, culminating in a highly stable 4-segment K-Means architecture validated through ablation testing and cross-tabulation.
R · Factor Analysis · K-Means · GMM · Hierarchical Clustering · Cross-tabulation
Hybrid RAG Architecture for World Bank Development Reports
2026
PythonFAISSBM25RRFCross-EncoderHyDELLM
Engineered a sophisticated Retrieval-Augmented Generation (RAG) pipeline to query over 3,000 pages of dense World Bank policy reports (2016–2025). Implemented FAISS dense vector search, BM25 sparse retrieval, Reciprocal Rank Fusion (RRF), Cross-Encoder reranking, and Hypothetical Document Embeddings (HyDE). Achieved a +0.40 lift in correctness and +0.10 in completeness over baseline dense retrieval evaluated via LLM-as-a-judge.
World Bank PDFs → Chunking → Dense (FAISS) + Sparse (BM25) → RRF Fusion → Cross-Encoder → LLM
Advanced Credit Risk Analytics & A/B Policy Simulation
2026
PythonA/B TestingK-MeansDeep LearningLendingClub
Formulated and simulated Conservative vs. Aggressive lending policies on a LendingClub dataset. Executed statistical A/B testing (Welch Two-Sample t-test, p < 2.2e-16), proving the growth-oriented policy generated +£298 profit per application despite higher default rates. Enhanced the risk assessment with an 11-feature K-Means clustering algorithm to uncover distinct borrower archetypes and modelled defaults using Deep Learning.
LendingClub Data → Policy A/B Simulation → K-Means Archetypes → Deep Learning Defaults → P&L Analysis
Predictive ML for Business Lending — A CRISP-DM Case Study
Apr – Aug 2026
PythonXGBoostLightGBMSVMCRISP-DMAUC
Built and benchmarked six ML classifiers (including XGBoost, LightGBM, SVM, and Random Forest) on a 100,000-record, 200-feature dataset to predict Universal Plus business credit applications. Addressed a severe 10/90 class imbalance and 40% missingness via median imputation, Information Gain feature selection, and embedded class-weighting, with XGBoost peaking at 0.861 AUC.
100K Records · 200 Features → Imputation → Feature Selection → 6 Classifiers → XGBoost 0.861 AUC
2026 FIFA World Cup Analytics: Optimized Scheduling & Group Formation
2026
PythonPyomoInteger ProgrammingPrescriptive Analytics
Redesigned the 48-team 2026 FIFA World Cup group stage and schedule using three integer programming models in Pyomo. Formulated a Max-Min Chebyshev objective to ensure group competitive balance, and solved stadium assignments with regulatory host-pinning constraints to maximize total match popularity across USA, Canada, and Mexico.
Data Prep → Max-Min Group Formation → Letter Assignment → Stadium Allocation
Relational Database Architecture & Analytics for Wealth Management
2026
SQL3NFChen ERDCrow’s FootETL
Designed and implemented an end-to-end 3NF relational database for a boutique UK wealth management firm. Developed conceptual (Chen) and logical (Crow’s Foot) Entity-Relationship diagrams. Executed complex SQL ETL pipelines to report on AUM growth waterfalls, branch-level unrealised gains, advisor portfolio risk metrics, and fee schedule revenue optimisation.
Conceptual (Chen) → Logical (Crow’s Foot) → 3NF Schema → SQL ETL → AUM · Risk · Revenue Reports
Runway to Rewards — WBS Business Challenge with The Emirates Group
Apr – Jun 2026
Business StrategyHook ModelGen ZLoyalty
Collaborated with Emirates Group executives on a live business challenge; led a 4-person winning team that earned a funded trip to Emirates HQ in Dubai. Designed Runway to Rewards, a gamified Skywards loyalty concept addressing the "360-day gap" between flights for Gen Z audiences using Nir Eyal’s Hook Model, daily micro-engagements, and a phased commercial rollout.
Emirates Challenge · 1st Place · Hook Model · Gen Z Loyalty · Gamified Micro-engagements