AI / Data Engineer · Freelance
AI telephone receptionist and lead-qualification, AI chatbots, automation and live reporting for regional SMBs.
Data Engineer — AI & Analytics
Jack Toke
I build reliable data pipelines & AI solutions that turn messy operational data into decisions.
Fig.01 — How I work: end-to-end data flow Batch + streaming
01
SOURCES
Calls · Ops · APIs
02
INGEST
Airflow · Beam · Pub/Sub
03
TRANSFORM
dbt · PySpark
04
WAREHOUSE
BigQuery · Delta
05
SERVE
Dash · Tableau
Databricks CertifiedQUT Grad Cert — Data AnalyticsLe WagonRMITBilingual Karen–English
01 / About
About
- Based in
- Ballarat region, Victoria
- Focus
- Data engineering + applied AI
- Languages
- English · Karen (bilingual)
I turn messy operational data into clean, trusted pipelines — and increasingly, into AI systems that act on it.
Data and IT professional with a Graduate Certificate in Data Analytics (QUT) and the Databricks Certified Data Engineer Associate credential. Hands-on across enterprise application support, systems integration, and business reporting.
At AME Systems I integrated Siemens Capital with enterprise systems and automated data synchronisation in Python & SQL. Now, working independently, I build AI and data engineering solutions for regional businesses — comfortable with stakeholders at every technical level.
02 / Skills & stack
Skills & tech stack
Data engineering
- SQL
- Python
- Spark / PySpark
- Databricks
- dbt
- Airflow
- Apache Beam
- Pub/Sub
Warehousing & ingestion
- BigQuery
- Postgres
- Delta Lake
- Auto Loader
- Fivetran
- Airbyte
Reporting & viz
- Tableau
- Plotly Dash
- Streamlit
Platform & other
- Docker & Compose
- Unity Catalog
- .NET (MAUI / ASP.NET Core)
- Kotlin / Jetpack Compose
03 / Experience
Experience
ICT Application Officer · AME Systems
Integrated Siemens Capital with enterprise systems (webhooks / SOAP); automated Python + SQL data synchronisation; built custom Sales and Engineering reporting; shipped a .NET MAUI + ASP.NET Core label app and Kotlin / Jetpack Compose Zebra scanner apps.
Computer Operator · Hazeldene's Chicken Farm
Ran and maintained computerised packaging and label machinery; kept production and breakdown records.
School Coordinator · Education Opportunity Foundation
Rolled out an English Teacher Toolkit across 20 schools on the Thai–Burmese border; trained 76 teachers; ran payroll for 20 schools.
04 / Projects
Selected work
01
Transport Victoria — GTFS-Realtime Pipeline
Production lakehouse measuring V/Line train punctuality from Public Transport Victoria’s GTFS feeds. A timer-triggered Azure Function (Python) downloads real-time protobuf trip updates to ADLS Gen2. A parallel Databricks job ingests the weekly static GTFS schedule across 11 bronze tables. Protobuf is decoded on the JVM with from_protobuf(), then modelled through a bronze–silver–gold medallion architecture with an SCD2 stops dimension. The gold fact flags on-time, severely late and skipped stops per served stop. Deployed serverless on Databricks with Unity Catalog and Bicep infrastructure-as-code, and served through a live Streamlit dashboard.
02
Australian real estate lakehouse & analytics dashboard
End-to-end Databricks lakehouse over Australian residential property listings, ingesting buy, rent and sold data from the realty-in-au API. Raw JSON lands via Auto Loader into a bronze–silver–gold medallion architecture on Delta Lake and Azure Data Lake Storage (ADLS Gen2). dbt transforms it into a dimensional star schema that separates physical properties from listings, with conformed dimensions, bridge tables and snapshot facts. Hash-based surrogate keys, referential-integrity checks and dbt data-quality tests (uniqueness, not-null, relationships) keep the data trustworthy. The property-versus-listing split enables repeat-sales analysis across 2007–2026, identifying 640 properties sold twice and 82 sold three times. Orchestrated with decoupled Databricks Jobs for nightly ingestion and file-arrival-triggered transformation, deployed as infrastructure-as-code with Databricks Asset Bundles across separate dev and prod catalogs, and served through an interactive Streamlit dashboard.
03
Citi Bike Medallion Pipeline
Event-driven lakehouse over roughly 10 years of Citi Bike trip data (2016–present). Auto Loader ingests raw CSVs and canonicalises their changing historical schemas into one format in testable pure Python. Change Data Feed streams typed, deduplicated trips into silver via MERGE upserts. The gold layer recomputes only the affected days per batch, staying correct under late-arriving data and deletes. Deterministic ride IDs are derived by SHA-256 hashing for records that lack native identifiers. Deployed on serverless compute with three GitHub Actions workflows for CI, gated deployment and job runs, using service principals across separate dev, test and prod workspaces. Served through a live Streamlit dashboard.
04
End-to-end ELT — Airflow, dbt & DuckDB
Dockerised ELT pipeline over JSON receipt data from four small businesses, running fully self-contained with no external services. Apache Airflow DAGs orchestrate ingestion and trigger downstream transformation. Modular dbt models transform the data in DuckDB. A Streamlit dashboard answers business questions on seasonal profitability, customer loyalty and employee turnover.
05
AI Telephone Receptionist & Lead Qualifier
Production AI voice system that answers inbound calls, qualifies and captures leads, and surfaces activity through live reporting for regional SMBs. Handles call routing, structured lead capture and automated follow-up, giving small businesses a first, practical step into AI adoption.
05 / Credentials
Certifications & education
Certifications
Databricks Certified Data Engineer Associate
Jul 2026
Data Engineering Bootcamp
Jun 2025
Data Engineering
Apr 2025
Education
Master of Biostatistics
In progress
Graduate Certificate of IT Practice (Data Analytics)
Dec 2025
Web Development Bootcamp
Feb 2020
BSc (Computer Science)
2010
06 / Contact
Get in touch
Let's build something reliable.
Open to data engineering roles and freelance projects. Drop a line and I'll get back to you.