Senior Analytics Engineer
Leading the data platform strategy across a high-volume marketplace — driving infrastructure ownership, FinOps, ML platform delivery, and an organisational shift from BI-focused analytics to a Core Data Platform serving ML, marketing, finance, and operations.
Highlights
- Authored the Data Platform Vision, directly driving the transition of the Analytics Engineering team to a dedicated Data Platform team.
- Delivered significant annualised cost savings across Snowflake, GCP, and vendor contracts — including a major contract renewal avoided by migrating audience activation to an existing platform tool, and implemented custom events pipelines to reduce dependency on a managed events layer, contributing to further contract savings.
- Reduced end-to-end dbt pipeline runtime from ~18 hours to under 3 hours across a large monolithic dbt project while also reducing failure rate, enabling next day analytics for the whole business.
- Delivered the Dagster orchestration platform end-to-end (vendor assessment → production K8s deployment), replacing fragmented cron scheduling with a single orchestration standard across Fivetran, dbt Cloud, and Snowflake.
- Designed and shipped Data Support Agent v2 — an AWS Bedrock-backed agentic orchestrator integrating Jira, Confluence, GitHub, Datadog, and Snowflake, with YAML-driven provisioning for new Slack bots.
- Led Snowflake security modernisation: enforced key-pair auth and Okta OAuth across all human and service users, cleaned up RBAC role grant sprawl, and defined a robust permissioning model separating business users, service accounts, and platform tooling.
- Implemented multi-region geopatial datasets that map to each regions statistical data. Enabling enrichment of marketplace performance with third party population data, driving emerging market detection, and guiding advertising decisions.
- Regularly involved in platform incidents and product development beyond the data team — leveraging deep knowledge of the product codebase and business domain to drive end-to-end data products back into the production monolith, with direct revenue impact.
- Built Airtasker's MLOps platform from scratch, enabling the team to ship real-time ML inference endpoints into a high-volume marketplace. Three models in production — task classification, offer ranking, and tasker interest matching — with a combined ~7% revenue uplift validated by Statsig A/B experiments.
- Collaborated with data team managers to redesign team ceremonies and knowledge-sharing during a re-org; runs weekly L&D / Office Hours and provides 1:1 mentoring to junior team members.