Data & AI Platform Engineer | WFH | Day Shift | Weekends Off

Location: Philippines (remote)
Engagement: Independent contractor, engaged directly by Pivot Wealth
Works with: Principal Engineer

About Pivot Wealth

Pivot Wealth is a Sydney based financial advice and mortgage broking firm. We help clients build wealth through advice that is genuinely personal, and we run a small, high accountability engineering team that builds the systems making that possible at scale. Our values are simple: the 1% matters, full transparency, results first.

About the engagement

This is a contract role engaged directly with Pivot Wealth. There is no BPO, agency or outsourcing provider in between: you contract with us directly, the full rate goes to you, and you work alongside the Pivot engineering team every day on Pivot systems. The Philippines runs two to three hours behind Sydney, so you will have strong daily overlap with the people you are building for.

What this role is

You’ll own the data platform behind an Australian financial advice firm. That covers the pipelines that pull client data, documents and calls into a Postgres client master on GCP, the Document AI and Gemini systems that read over 90,000 client files, and the LLM tools that sit on top.

The problem we most want solved

We classify client documents into 58 types. Around half never get classified at all, and we cannot currently tell you how accurate the extraction is. There is no ground truth set, and human review decisions are sitting unused as labelled data. Making that measurable, then making it better, is the highest value work in the business, and it is yours.

What you’ll own

Document extraction: Document AI custom classifiers and parsers, Gemini extraction, confidence gating and human review queues

The client master: a Postgres gold layer with per field provenance, confidence and human overrides

Ingestion: HubSpot, Box and Zoom sync, webhooks, reconciliation and backfills

LLM systems: a rubric anchored judge scoring adviser calls, a compliance scanner, and an MCP server exposing client data to internal assistants

The surface: internal Flask apps and APIs that put extracted data in front of a human to check

Your first 90 days

Land a backlog of unmerged fixes, build a labelled ground truth set and the first accuracy measurement, separate staging from production, and add alerting as code.

You’ll need

5+ years of Python, building ingestion from third party APIs rather than configuring connectors

Deep PostgreSQL: large views, JSONB, migrations, roles and column level grants

Production GCP: Cloud Run services and jobs, Cloud SQL, Pub/Sub, Scheduler, Secret Manager, IAM and some Terraform

At least one AI or ML feature you shipped to production and can describe being measured, ideally document extraction or classification at scale

Experience being the person responsible when a pipeline broke, and writing the runbook afterwards

Experience taking over a substantial codebase after its author had already left

Human in the loop design: confidence thresholds, review queues, and override layers that keep machine values and human corrections distinct

Scoped, fail closed read APIs over sensitive data: parameterised queries, identity verified server side, and access enforced in SQL

Nice to have

MCP servers, weak supervision or noisy label techniques, annotation tooling, and Australian financial services domain knowledge.

This role is not

A warehouse and DBT specialist, a big data engineer, or an ML researcher. There is no Airflow, no Snowflake and no Spark here, and there won’t be.

Working with sensitive data

You will be handling Australian client financial data. Privacy, access control and auditability are not afterthoughts here. They are part of how the systems are designed from the first line of code.

Why this role

You get direct access to the Sydney team and the business problems behind the work, not a ticket queue. The systems you build are used every day by the advisers and brokers serving our clients, so you see the impact of your work quickly.

How we hire

Our process is short and practical. It starts with an initial screening call, followed by a technical skills test built around the kind of pipeline and extraction work you’d be doing here, then a final interview with our engineering leadership. We care far more about what you’ve built and owned than about where you’ve worked.

Unlock job insights

Hirer responsiveness Salary match Number of applicants