Projects

Projects & Experiments

A mix of customer data, AdTech, cloud, AI, and curiosity-driven builds.
Now · Amperity
Customer data architecture

As a Senior Solutions Architect at Amperity, I design how enterprise brands bring their customer data together on an AI-powered customer data platform.

  • Data discovery and integration design across source systems, cloud warehouses, and marketing platforms.
  • Identity resolution and a unified customer 360 that teams can trust.
  • Activation strategy: getting the right customer data to every channel that needs it.
Lab
Interactive explainers

Small things you can poke at, for ideas I work with every day. No sign-up, no real data.

Side project
Agent X — conversational clean-rooms analytics

Plain-English questions in, privacy-safe SQL out. A conversational analytics layer over AWS Clean Rooms and Amazon Marketing Cloud — built for advertisers who'd rather ask "what's my TikTok + Amazon DSP overlap?" than write 200 lines of restricted SQL.

  • AWS Bedrock (Claude Sonnet) + RAG over Clean Rooms / AMC schema knowledge.
  • Self-improving few-shot — successful query→SQL pairs re-inject into the prompt, so the model gets better with each session.
  • 11 data partners connected via IAM roles and OAuth (AMC, TikTok, Trade Desk, AdSquare, LiveRamp, Salesforce, …).
  • Privacy-safe by default: every generated SQL statement is sanitized before execution; no raw PII touches the model.

Explore the platform →

Amazon Ads · 2021 – 2026
Measurement & modelling

I’ve led multiple initiatives around custom modelling on Amazon Marketing Cloud and AWS Clean Rooms, building MMM, MTA, audience, and bid models for global brands.

  • Designed privacy-aware data collaboration workflows for advertisers.
  • Built modelling environments that could be reused across campaigns.
  • Turned model outputs into simple, usable insights for teams.
Side projects
Smaller experiments

Outside work, I like building small tools or prototypes that make life a bit easier or help me learn something new.

  • A private family health portal: AI reads uploaded medical reports, a human reviews, and the family gets a daily digest of what's due.
  • Data visualisations to explore cricket or fitness stats.
  • Little scripts and automations to reduce repetitive tasks and noise.
  • Experiments with new cloud or AI services, just to see what’s possible.