// background
I'm an Analytics Engineer specialising in revenue and GTM data for B2B SaaS. My work sits at the intersection of data engineering, analytics, and business intelligence — building the warehouse layer that finance, CS, and product teams rely on daily.
Across my career in B2B SaaS I've led CRM unification projects, rebuilt revenue data pipelines from the ground up, and shipped machine learning models into production CS workflows. I've owned Snowflake, dbt, Salesforce, and Tableau end-to-end across the whole revenue data lifecycle.
My current focus is integrating AI as a governed layer on top of warehouse assets — not replacing analytical rigour, but amplifying it with safer, auditable LLM tooling.
// technology
Warehouse
Transformation
Languages
BI & Viz
Orchestration
Sources & AI
// open source work
Governance-first AI demo for Customer Success teams. LLM queries are restricted to a curated dbt allowlist — parameterised templates only, no free-form SQL generation. Mirrors how I'd build this in production.
Production-grade A/B testing pipeline. Assignment and exposure tracked as separate events, daily exposure quality monitoring, explicit ITT vs exposure-based cohort selection. Statistical output only after integrity checks pass.
Live anomaly detection on ADS-B aircraft data over 4 European cities. Demonstrates the same observe → baseline → compare → flag → explain pattern used in production revenue monitoring systems.
Behavioural event pipeline from raw stream to governed self-service layer. Clean staging → intermediate → mart separation prevents raw event schemas from bleeding into BI tools.
// production impact
| Domain | Impact | Context |
|---|---|---|
| Revenue accuracy | 12–15% customer overlap found 8–10% revenue overstatement corrected | CRM data reconciliation across two product lines following a business integration. Identified customer overlap and corrected revenue double-counting in subscription data. |
| Pipeline performance | 78 min → 7 min 91% reduction in CARR pipeline runtime | Full rebuild of a subscription revenue pipeline — incremental materialisation, query restructuring, and partitioning strategy. Daily refresh completed well within business hours. |
| Churn ML model | 86% accuracy · AUC-ROC 0.91 ~5% churn reduction in target segment | XGBoost churn risk model built on warehouse-sourced features and deployed to production. Scored weekly per account; integrated into customer success tooling. |
| Data migration | 100k+ records migrated Zero data loss, zero downtime | Subscription data migration during a billing platform consolidation. Full validation suite run against source and destination before cutover; zero rollbacks required. |
| Reporting efficiency | ~40% reduction in manual reporting time | Standardised dbt models replacing ad-hoc analyst SQL across the reporting layer. Executive dashboards and recurring business reviews moved to self-serve certified extracts. |
Designed and implemented a unified revenue data platform that consolidated fragmented reporting logic into one trusted framework — enabling Finance, Revenue Operations, and leadership to work from consistent metric definitions.
Built a curated analytics layer that allowed business users to answer complex questions independently, without ad-hoc data requests. Improved accessibility, consistency, and governance of key business metrics across the organisation.
Led data quality, reconciliation, and metric standardisation initiatives that moved organisations from conflicting reports and manual validation toward trusted, decision-ready data.
Designed scalable data models and transformation frameworks capable of supporting business growth while maintaining transparency, auditability, and reliability across the analytics stack.
Developed customer analytics frameworks enabling teams to identify risk patterns, prioritise interventions, and shift from reactive to proactive business decisions.
Introduced product-oriented approaches to analytics: ownership models, documentation standards, monitoring practices, and reusable assets — reducing operational complexity and improving adoption.
// current focus
Building governed AI layers on top of warehouse assets — intent routing, parameterised SQL templates, output validation.
Cross-project dependencies, model contracts, and the dbt semantic layer — unified metric definitions queryable across tools without duplicating logic.
Native ML functions, Cortex Analyst, and document AI for embedding AI directly into the warehouse layer.
Principal-level skills: team enablement, data contracts as organisational boundaries, and analytics-as-a-product thinking.
// get in touch