The roleYou'll own Passage's data platform end to end. Not a greenfield project and not a legacy mess: a working, documented warehouse that the company already runs on - executive KPIs, marketing attribution, funnel analytics, and internal ops tools are all powered by it today. Your job is to take ownership, keep it trustworthy, and extend it as the business grows.
This is a high-leverage, high-trust seat. The models you build define how the company measures itself, and your stakeholders are leadership, marketing, and operations - not a ticket queue.
What you'll do- Model the business in dbt on BigQuery. Own a ~100-model project (staging → intermediate → marts) built on Kimball dimensional modeling: SCD Type 2 dimensions, fact tables with explicit grain, and aggregate tables for reporting. Recent work includes applicant milestone tracking, loan-process stage modeling, and ad-click-level attribution.
- Own metrics people rely on. Executive KPIs, conversion funnels, cohort analytics, and marketing attribution across Google Ads, Meta Ads, and UTM/click-ID data.
- Run the pipelines. Airflow 3 orchestration with Slack alerting, data-quality tests, freshness monitoring, and backfills. Ingestion is CDC from our production Postgres via Debezium + Kafka Connect on GKE (with Airbyte and Fivetran alongside).
- Serve the data. A dedicated dbt serving layer exposed through Hasura GraphQL powers internal ops tools in the product itself; Metabase serves dashboards and self-serve queries.
- Turn messy operational reality into tables people trust. Much of the work is business-logic modeling: screening outcomes, payment and fee states, deferral lineage, guarantor chains. You'll dig into the source system, talk to the people who run the process, and encode the truth.
- Keep quality and cost in check. dbt tests, per-model documentation, BigQuery partitioning/clustering, and bytes-scanned discipline.
- Document for humans and AI. Every model is documented, and the docs auto-sync to a companion repo built so both teammates and AI agents can query the warehouse correctly. We work AI-assisted, with coding standards written into the repos.
Our stackBigQuery • dbt • Airflow 3 • Debezium + Kafka Connect (GKE) • Airbyte • Fivetran • Hasura • Metabase • PostgreSQL (Django app) • Docker • GCP • AI-assisted development
You might be a great fit if you- Have 4+ years building analytics or data platforms, with expert-level SQL.
- Have run dbt in production and have real dimensional-modeling instincts - you think in grain, and SCD Type 2 doesn't scare you.
- Have operated an orchestrator (Airflow or similar) and are comfortable in Python.
- Enjoy ambiguous business-logic modeling: reverse-engineering a production schema, interviewing stakeholders, and shipping a table that settles the debate.
- Write documentation people actually use.
- Can own a platform solo: prioritize, communicate, and ship without a spec.
Nice to have- CDC pipelines (Debezium, Airbyte) or Kafka Connect; Kubernetes/Helm exposure.
- BigQuery performance and cost tuning.
- Hasura/GraphQL, reverse ETL, or embedding warehouse data into product surfaces.
- Metabase or Looker administration.
- Experience as an early or solo data hire at a startup.
- Fluency with AI-assisted development workflows.
What success looks like in the first six months- You run the platform day to day - pipelines, tests, and monitoring - and stakeholders don't feel a handoff.
- You've shipped new marts or metrics for at least two teams, from stakeholder conversation to documented, tested tables.
- Data quality and freshness are visibly better: fewer surprises, faster answers.
Compensation
C$100K to C$150K + Offers Equity
