Consultancy

A small number of bounded engagements alongside my in-house role. Two modes, depending on what you need:

• Build: hands-on delivery — I'll stand up your warehouse, BI stack, or analytics pipeline myself. Code, infra, dashboards, the lot.
• Advise: strategic input for founders, CTOs, or boards — architecture review, hiring, due diligence, separating AI hype from leverage.
• Scope: NDA-friendly, non-compete with my employer's sectors (beauty, tattoo, pet grooming supply). Anything else is fair game.
• Intro: Mostly via referral — if you've landed here directly, mention who pointed you my way. Async by default; calls evenings or weekends.

Grounded in what I've actually built, not theory. These are the areas where I can compress years of trial-and-error into weeks of advice.

• Data system design from scratch: end-to-end architecture for companies with no data function yet — what to build, in what order, what to skip. Hourly syncs from NetSuite, Shopify, Google, and friends into a centralised warehouse using Python, SQL, Polars, DuckDB, and Docker.
• BI & reporting from zero: replacing legacy spreadsheets and disconnected tools with a unified Streamlit (or equivalent) stack — SSO, role-based access, dashboards covering exec, ops, finance, and marketing. Pragmatic alternative to per-seat BI licences.
• Analytics & commercial modelling: CAC modelling and optimisation, pricing & shipping margin analysis with statistical significance testing, basket & bundle intelligence, discount dependency, vendor spend, inventory and OTIF diagnostics, EOS L10 scorecards.
• AI enablement: embedding LLM-assisted development (Claude Code) into data and engineering teams to compress pipeline and dashboard delivery. Helping leaders separate hype from where AI actually moves the needle.
• Self-hosted local LLMs: standing up private, zero-marginal-cost models for sensitive workloads — review sentiment, SEO copy analysis, product description harmonisation, classification — with prompts tuned per task. No data leaves your network.
• Hiring & team build: what a first data hire should look like, when to add the second, how to structure a small high-leverage team. Drawn from building and leading a team of PhD-level scientists and engineers.
• Technical due diligence: independent reviews of data & AI stacks for investors, acquirers, or boards considering an investment.

Nexus Brands Group. I built the data function from scratch across 18 businesses in beauty, tattoo, and pet grooming supply — warehouse, BI, analytics, AI, and the team to run it. Selected outcomes:

• +25% OTIF improvement from unified inventory and order visibility.
• ~1% of revenue recovered that was previously lost to out-of-stock.
• £Xm in excess & obsolete inventory identified and surfaced to leadership (figure withheld).
• Designed and built FLARE — a self-hosted warehouse + dashboard platform replacing legacy reporting across the group. Zero per-seat cost, every employee has access. Anonymised public demo →
• Built and lead a team of PhD-level data scientists and engineers supporting senior leadership across the group.
• Evidence base underpinning a major global restructuring and optimisation programme.