ADRIAN WHITE
APPLIED AI & DATA ENGINEERINGAUCKLAND, NEW ZEALAND

AdrianWhite.

Turning complex data
into working systems.

Hands-on AI engineering, data architecture and technical leadership. More than 20 years of making difficult data useful.

Explore selected work
Python · SQL · LLMs · Data platformsBased in Auckland · Open to Sydney

01 / SELECTED WORK

Build it. Inspect it.
Make it reliable.

Applied AI grounded in data modelling, validation and software you can examine.

FEATURED PROJECT / SOLE BUILDER

SNOWBALL

An LLM-assisted data warehouse builder

Designed and built end to end. A company is currently trialling SNOWBALL with its ERP system.

Profiles source data, infers relationships and proposes dimensional models. Generated SQL, schema designs and source-to-warehouse lineage stay reviewable.

Built with Python, PostgreSQL, FastAPI and React, with dbt models and tests, validation and reconciliation, and CI checks across the backend, API contracts and browser workflows.

PythonPostgreSQLLLM APIsdbtFastAPIReact
Open SNOWBALLSign-in required
FROM SOURCE TO WAREHOUSESNOWBALL / 01
01

Understand the source

Profiles · keys · relationships

02

Propose a model

LLM-assisted dimensional design

03

Build and validate

dbt · SQL · tests · reconciliation

REVIEWABLE DESIGN, CODE AND LINEAGE

Workflow diagram · App preview requires sign-in

More projects

Websites and applications built by Adrian White.

TRAINING & AI

The Crowd

An app designed to connect Garmin training data with AI coaching and training-load views. Training features are behind sign-in.

  • 3D route replay over photorealistic city tiles
  • Fitness, fatigue and form over time
  • Garmin activity data connected to AI coaching
Visit The CrowdSign-in required for training features
The Crowd 3D flyby replay of Adrian’s Tokyo running route over photorealistic city tiles, with map attribution preserved
01
Replay the route.Tokyo · 3D flyby
The Crowd fitness, fatigue and form chart with the original August-to-October time axis and October 4 tooltip values
02
Inspect the training.Fitness · fatigue · form
Snadly public landing page introducing its AI-assisted data warehouse platform
Public landing page

AI & DATA WAREHOUSING

Snadly

A public product site for an AI-assisted data warehouse platform, with generated dashboards, source-to-warehouse lineage and inspectable relationship inference.

Visit SnadlyPublic product site
Umeda Sky Building reconstruction with the completed towers and construction timeline
Interactive 3D demo

INTERACTIVE 3D

Umeda — Building the Sky

A 3D reconstruction of Osaka’s Umeda Sky Building, with eight construction chapters, a timeline and interactive camera views. Modelled in Blender.

Explore Umeda SkyPublic 3D reconstruction
Prompt Instead guide library with search and workflow categories
Public guide interface

AI WORKFLOWS

Prompt Instead

A searchable collection of practical AI prompt guides, with company-specific workflows, limitations and a subscription cost-comparison calculator.

Visit Prompt InsteadPublic project
Brick Explorer displaying the Millennium Falcon model and model controls
Interactive 3D demo

WEEKEND BUILD

Brick Explorer

A weekend build for exploring LEGO sets, selected animated 3D builds and parts. The catalogue and explorer are available without an account.

Visit Brick ExplorerNo account needed
LexiLabs message editor and language selection interface with an empty input
Public writing interface

LANGUAGE & WRITING

LexiLabs

A writing tool with a public interface for rewriting messages, choosing a language and adjusting tone.

Visit LexiLabsPublic writing interface
Huff model video still showing supermarket markers and coloured catchment areas across central Auckland
Video walkthrough · 0:49

SPATIAL MODELLING

Huff Modelling

A supermarket catchment modelling demo using Huff probabilities, choropleth maps and interactive sliders. Explore candidate sites and potential cannibalisation across a 605-store dataset.

Visit Huff ModellingWatch walkthroughPassword protected

02 / CONTRIBUTIONS

Applied AI.
Shared knowledge.

AI integrations, model red-teaming, prompt engineering and technical contributions to the data community.

FEB 2023

World-first OpenAI / Snowflake demo

Delivered the world’s first public demo of an OpenAI integration with Snowflake at the Auckland Snowflake User Group in February 2023.

Read the technical write-up
PUBLIC DEMO
MAY 2025

Anthropic Model Safety Bug Bounty

Participated in the programme and undertook model red-teaming tasks.

PARTICIPANT
OCT 2026

Prompt Golf

Ranked first on each of the six hole leaderboards as Par, verified 2 October 2026.

View the leaderboard
6 HOLES / 6 FIRSTS
2023 / 2025

HackAPrompt

Competed in prompt-injection challenges, including Level 8 in 2023 and the 2025 CBRNE track.

COMPETITOR
JUN – AUG 2022

Stack Overflow · Snowflake

Number-one contributor for the Snowflake topic on Stack Overflow from June to August 2022.

View Stack Overflow profile
#1 CONTRIBUTOR

03 / EXPERIENCE

Engineering depth.
Business context.

From data warehouses and cloud migrations to multimodal extraction and AI-enabled workflows.

Flowingly

Sep 2025 – Present

Full Stack Developer

Workflow and standard operating procedure features using SQL Server and Azure AI Foundry. Prototyped a Python integration with the Naverisk API and validated basic lookup functions.

Infinity Studio

Jun 2024 – Feb 2025

AI and Data Engineer

Used Python and multimodal models to extract data from legacy geotechnical reports, including hand-drawn graphs. Enhanced a Databricks warehouse for billing accuracy and data processing.

ArchiPro

Oct 2023 – Jun 2024

Lead AI and Data Engineer

Delivered a working Snowflake data mart in four weeks, bringing together Google Analytics, website logs and CRM data. Guided newly hired engineers after the initial delivery.

Kiwibank

May 2022 – Oct 2022

Data Modeler

Integrated public datasets into Snowflake for property-market analysis, identified source discrepancies and developed ETL lineage solutions for data governance.

Sky TV

Mar 2021 – May 2022

Data Modeler

Designed Snowflake models and automated validation. Reduced a query across 1.5 billion television records from 36 hours to under six minutes by changing its join predicate.

Read the SQL optimisation write-up

Countdown

Nov 2018 – Jul 2020

Data Solutions Engineer

Worked alongside Google engineers on a Teradata-to-BigQuery migration, unified online and offline customer data, and optimised data loads with BigQuery and Airflow.

Icebreaker

Jun 2013 – Nov 2018

Senior Data Warehouse Architect

Redesigned data models around core business processes, reducing load times from 14 hours to 75 minutes. Built fault-tolerant ETL with SQL and Node.js.

FOUNDATIONS

Earlier data warehouse work at Spark NZ TNZI, Telstra Wholesale and Vodafone.

TRAINING & CONTRIBUTION

Google Cloud Professional Data Engineer certification · dbt Fundamentals · Dimensional modelling training with Ralph Kimball

Stack Overflow contributions