graph by datavruti
mygraph.id/haydnflower67

Haydn Flower

Systems Analyst / Data Engineer

Perth WABuilding since 2001 (~26 yrs)
Headline

Across EDI, SOAP, REST, and now AI-assisted workflows, I consistently apply the same evidence-first method: profile real data, find the gap, fix it safely, then verify against a known-good result.

225Data Quality Rate

of 228 validation errors cleared in one regulatory batch

Represented Roles

Specialized Functions & Roles.

Engineering capabilities and specialized roles demonstrated across real production work.

Data Engineer

Primary Role

Systems/Integration Analyst

ETL Developer

BI Developer

AI-Assisted Engineering Specialist

Signals

Live in Production.

Each signal ties back to a specific, evidenced moment.

Solved

2 moments

Gaps closed - false positives, edge cases, dead ends.

  • •Enforced evidence-only root-cause investigation on production data
  • •Cleared a regulatory validation batch down to near-zero errors

Learned

1 moment

New tools and depth picked up on the job.

  • •Experimenting with Claude Code to build a LinkedIn posting cadence

Optimized

1 moment

Made faster or cheaper: runtime, latency, spend.

  • •Cut AI session context load by over 95 percent
Growth trajectory

Haydn's career, mapped.

A timeline of milestones captured directly from real work.

  1. Started building cross-protocol data integrations

  2. Moved into process and requirements ownership

  3. Owned a live system cutover

  4. Scaled into enterprise data warehouse work

  5. Took ownership of a production ingestion layer

  6. Began engineering around AI-assisted workflows

  7. Formalized evidence-only root-cause investigation

Stack

Daily tools.

What Haydn actually reaches for, pulled from the tags attached to captured moments.

Work

Key Work By Role

What they built, shipped and owned at each role, with impact numbers where there are any.

HSI Donesafe

Data Wizard · Jul 2017 - Present

Enforced evidence-only root-cause investigation on production data

Ran AI-assisted root-cause investigations against live production databases under a strict rule: no cause is stated as fact until a query result or line of source code supports it, with everything else logged as a hypothesis plus the evidence that would confirm or kill it.

Traced a silent import failure caused by a single escaped character that had gone undiagnosed for months.
Stack:AI-assisted engineeringSQLRCAProduction debugging
View case study

Cleared a regulatory validation batch down to near-zero errors

Remediated flagged regulatory data safely using a transaction-wrapped, rollback-first process previewed with before-and-after row counts and checked against regenerated output before anything committed.

Cleared 225 of 228 validation errors in one regulatory batch.Confirmed a 1,827-payment replacement file matched the expected patch preview.
Stack:Data remediationSQLRegulatory compliancePython
View case study

Cut AI session context load by over 95 percent

Turned each failure of the AI workflow into an enforced control, including a hook making schema checks mandatory, rules for personal-data handling and evidence standards, and written handoffs so context carries between sessions.

Cut standing AI context per session from about 40k tokens to about 1.5k by indexing the skill catalogue and loading detail on demand.
Stack:AI-assisted engineeringClaude CodeWorkflow design
View case study

More from this role

  • Design and maintain the data ingestion/extraction layer for a production web application (AWS + Azure SQL) used for Injury and Safety and Return to Work management.
  • As part of the Implementation Team, built the reproducible ETL/ELT process that ingests new client data during onboarding. Source files arrive via FTP and email in inconsistent formats, and I validate, transform and load them reliably with Python and SSIS regardless of format drift.
  • Support that pipeline in production, troubleshooting data-quality errors as they surface.

Ministry of Business, Innovation and Employment (NZ)

Technical Business Analyst · Oct 2012 - Jul 2015
  • Part of the Business Intelligence team supporting the NZ Companies Office (15+ registers, each with its own schema, plus a linked payments/accounts database), using SAP Business Objects, Oracle SQL and Crystal Reports.
  • Reconciled PPSR (Personal Property Securities Register) payments against registration volumes to find renewal-payment discrepancies.
  • Built an automated monthly invoice-creation process for the direct-debit system, saving person-days of manual work every month.

SCM Group

Information Manager · Apr 2011 - Mar 2012
  • Designed and maintained a MySQL database (with replication) to process client data, and a high-availability system with disaster-recovery failover.
  • Ingested EIEP1 electricity data files into MySQL with Python and charted usage over time in Google Sheets to spot anomalies, as part of power reconciliation work on energy time-series data.
  • Implemented website design in Silverstripe CMS.

Air New Zealand

Technical Business Analyst · Sep 2010 - Mar 2011
  • Strong focus on system integration and interfaces for a new cargo solution, working with existing Air NZ systems.
  • Monitored and ensured delivery of all messages in and out of the new and incumbent systems during testing and go-live.
  • Worked with XML SITA messages, IATA C-IMP messaging types, and Type B protocols.

ETITO

Business Analyst · Nov 2009 - Dec 2009
  • Process-oriented data analysis, system and process development, testing and documentation.
  • Entered business processes into Mavim Rules software; data warehouse validation.

Zeald.com

Technical Business Analyst / Customisations Coordinator · Jul 2006 - Apr 2009
  • Ran project reporting, budgeting and resource scheduling across the PMBOK process groups for multiple simultaneous projects.
  • Interviewed business owners to clarify requirements and turned them into functional specs, primarily via use-case scenarios.
  • Instigated changes to the customisation scheduling and communication process that measurably increased customer satisfaction.

Spargonet

Consultant · 2004 - 2006
  • Data, ETL and integration consulting.

Kewill

Developer · 2001 - 2004
  • Data, ETL and integration work across EDIFACT, ANSI X12, XSLT, XML, FTP, HTTP, AS2, WSDL and SOAP architectures and protocols.
Outside Work

Side projects, open source & publications.

Independent builds, open-source contributions, technical writing, and experiments beyond the day job.

Experiment

LinkedIn Content Cadence

Playing with Claude Code to build a LinkedIn content posting cadence, aiming to grow an audience.

Present
Experiment

Local self-hosted LLM setup

Set up a local, self-hosted LLM on existing hardware to handle high-volume low-value work at zero cost, reserving paid AI reasoning for tasks that actually need it.

Side project

Self-hosted content-analytics dashboard

Built a self-hosted content-analytics dashboard; caught and fixed a silent bug where a 'successful' script run had quietly written placeholder data instead of real numbers, three separate times with no error raised, then instrumented it so it can't recur unnoticed.

Learning

Formal record.

  • Diploma, Information Technology (with Distinction)
    Seek Learning
    2009
  • CAPM (Certified Associate in Project Management)
    PMI
    2009
  • Diploma of Programming (Applications)
    Auckland University of Technology
    2000
Activity heatmap

A record of momentum.

A combined view of Haydn's activity across the tools used daily.

LessMore

Profile last updated today.

Liked Haydn's graph?

Build yours today.

Start with your resume