You will own the customer-facing data platform behind ProfitPeak's core product: a Databricks lakehouse that turns raw commerce, inventory, marketing and finance data into order attribution, revenue intelligence and marketing analytics data products for retail brands and the AI agents that act on their behalf.
This is a hands-on engineering role with product ownership. You decide what the platform ships and in what order, you build the pipelines and models that ship it, and you are accountable for the freshness, correctness and reliability of every number a merchant or an agent sees.
About ProfitPeak
ProfitPeak is an agentic platform for global retail brands. We ingest data from every commerce, marketing and finance platform a brand runs on, model it into a single view of profit, and put AI agents to work on it through our product suite.
We are headquartered in Sydney with an engineering team in Surry Hills and Malaysia, and we are expanding into the US.
Own the platform
- Architect and run the customer-facing data platform on Databricks (Lakeflow, Delta Lake, Unity Catalog) as infrastructure-as-code, with PII masking, access controls and governance built in from the start.
- Design and operate the serving layer on ClickHouse so that product queries and agent calls return in milliseconds against multi-tenant data.
- Own the roadmap for our data products: order attribution, revenue intelligence and marketing analytics. Decide what ships, in what order, and define what "correct" means for each.
Build the data products
- Build and orchestrate multi-tenant revenue and attribution models (multi-touch attribution, FX conversion, cost allocation) with Lakeflow Jobs and Spark Declarative Pipelines, guarded by SDP expectations and Unity Catalog constraints.
- Productionise forecasting and pricing models (order forecasting, price regression) with MLflow, Model Registry and Lakeflow Jobs, including scheduled retraining and scoring pipelines that ship predictions directly into the product.
- Expose data products as clean, versioned interfaces that both the Nexus UI and Sherpa agents consume, and ship the data outputs that agentic workflows act on: Watch (signals and anomalies), Advise (recommendations with the numbers behind them) and Act (tool‑ready datasets agents write back to Shopify, Klaviyo and ad platforms).
Ingest and contract
- Implement streaming ingestion for every merchant integration (Shopify, Klaviyo, Google, Meta and TikTok Ads) over AWS Kinesis.
- Design cross-team data contracts with freshness and completeness monitoring, so that upstream changes never silently break downstream numbers.
- Lead the ERP integration project: design and build the inbound API that lets customers push inventory, cost and order data from NetSuite and AP21 into ProfitPeak, with a freshness target of under 4 hours.
Ship safely and often
- Build the CI/CD and local development workflow that validates every change against production-shaped data before merge, so the team deploys frequently with zero customer‑facing data regressions.
- Set and hold SLOs for data freshness, completeness and accuracy, and own the on‑call and incident process for the platform.
- Work directly with customers, Customer Success and Sales to understand how brands use our numbers, and turn that into platform priorities.
What you’ll bring
- Experience building production data platforms on Databricks: Delta Lake, Unity Catalog, Lakeflow (Jobs and Declarative Pipelines / DLT), Spark SQL and PySpark.
- Experience shipping multi-tenant, customer‑facing data products where correctness is the product, not an internal dashboard.
- Streaming ingestion in production (AWS Kinesis, Kafka or equivalent), including schema evolution, late and duplicate events, and backfills.
- MLOps experience with MLflow and Model Registry: taking a model from a notebook to a scheduled, monitored retraining and scoring pipeline.
- Fluency in data modelling for commerce: orders, refunds, cost of goods, ad spend, attribution windows, FX and currency handling.
- Infrastructure-as-code (Terraform or Databricks Asset Bundles) and CI/CD for data, including testing against production-shaped data.
- Hands-on with a low-latency serving store (ClickHouse strongly preferred; Pinot, Druid or similar acceptable).
- A product owner's mindset: you write the spec, prioritise the backlog, say no to the wrong work, and measure the outcome.
- Clear written communication; you can explain an attribution model to a merchant and a data contract to an engineer.
- Based in Sydney, or willing to relocate; full working rights in Australia.
- Your work is the product. Every pipeline you ship changes the numbers brands run their business on and the actions AI agents take on their behalf.
- Modern stack, no legacy: Databricks, ClickHouse, Kinesis, MLflow and infrastructure-as-code, relaunched in 2026 and still being shaped.
- Real ownership. You set the roadmap for the data platform and work directly with founders, customers and the engineering team in Sydney and Malaysia.
- Growth ahead: we are expanding into the US market, and the platform you build is what we take there.
- Market salary, generous ESOP and an annual global offsite (2026 was in Bali).