38
Documented failure drills shipped
38
Documented failure drills shipped
6
Median facilitator response (hours)
11
Cities represented in last cohort
62%
Repeat partner teams
214
Office hours hosted (rolling year)
No. We map tradeoffs across Airflow, Dagster, Prefect, Luigi, and cloud-native schedulers so teams can sequence migrations honestly.
Foundations tracks welcome analysts moving into pipeline ownership; Advanced tracks assume comfort with SQL, Git, and basic cloud consoles.
See the Refund & intake rules page for eligibility, timelines, and non-refundable items specific to live cohorts.
Core labs are live with recordings for missed segments; some briefings are async by design to respect global time zones.
Rolling practicum hours
127
Hours of live pipeline critique delivered last quarter across Airflow, Dagster, Prefect, and warehouse tracks—no hidden “magic autopilot” modules.
Contract-first marts with reviewer-ready diffs.
Dual-write drills with comms templates attached.
Urbadan hosts small-group engineering studios in Singapore for teams who want warehouse and orchestration changes to feel legible—not heroic.
Every track ends with tangible templates: runbooks, alert copy decks, schema review agendas, and finance-friendly slot narratives tuned for APAC leadership rhythms.
Hands-on patterns for landing raw telemetry, enforcing contracts, and promoting datasets with observable quality gates.
→Design SLAs, lineage-aware retries, and pool tuning so nightly loads recover without pager storms.
→Model pipelines as assets with explicit dependencies so analytics and ML consumers read the same graph.
→Five beats, horizontal on wide screens, stacked on phones—never four pastel circles pretending to be “strategy.”
Diagnose orchestration habits with anonymized DAG samples.
Rehearse cutovers using tabletop timelines instead of slide decks.
Publish runbooks product and finance teammates can skim.
Instrument alerts with human sentences, not color codes alone.
Graduate with artifacts pinned in your internal handbook.
Thirteen live tracks spanning orchestrators, warehouse tooling, and governance rituals. Adjust filters to narrow the grid—desktop shows sidebar, mobile stacks controls first.
Hands-on patterns for landing raw telemetry, enforcing contracts, and promoting datasets with observable quality gates.
Design SLAs, lineage-aware retries, and pool tuning so nightly loads recover without pager storms.
Model pipelines as assets with explicit dependencies so analytics and ML consumers read the same graph.
Compose resilient flows with dynamic tasks, caching, and human-in-the-loop approvals for regulated batches.
Blend Kafka-style streams with Iceberg tables, focusing on idempotent writers and compaction windows.
Refactor long-running Luigi graphs with clearer targets, dependency visualization, and failure isolation.
Ship slim CI, defer syntax, and state-aware builds so pull requests stay fast even on wide tables.
Author Dataform pipelines with assertions, row access policies, and BigQuery slot awareness baked in.
Chain ADF activities with meaningful monitoring, parameterized datasets, and controlled fan-out.
Apply Great Expectations-style checks plus producer-consumer contracts before promoting schemas.
Score legacy cron sprawl, inventory dependencies, and sequence a low-risk migration to a modern scheduler.
Allocate warehouse spend to products using dbt exposures, query history, and finance-friendly dashboards.
Treat lineage diagrams, README templates, and incident notes as first-class artifacts for new hires.
4.8 / 5 verified studio note
“The Bronze-to-Silver Modeling Lab forced us to name every assumption in our staging models. Our audit trail finally reads like a story instead of a pile of SQL files.”
“Airflow SLAs Without Firefighting reframed how we write pages. I still keep the capstone runbook pinned above my desk—not because incidents happen nightly, but because it reminds the team what calm looks like.”
“Prefect cohort was short, but the human-in-the-loop approvals lab mirrored how legal actually wants to sign batches.”
“Client in logistics: appreciated the honesty about hybrid orchestration costs; no magical autoscaler promises, just staged waves.”
“Streaming Lakehouse Cutover Studio was dense, and I would have liked one more office hour on compaction math—but the failure drills alone justified the time away from my backlog.”
Trusted by platform pods at Horizon Freight, Lighthouse Health, and Nimbus Billing for facilitator-led critiques. Internal pulse surveys after 2025 cohorts: 94% of squads said rituals felt sustainable—not extractive.
How we coach teams to narrate slot usage as a product metric instead of a scolding exercise.
Read articleTurning DAG graphs into guided tours for newcomers joining mid-quarter.
Read articleRegistration closes 30 June 2026 for Singapore-facilitated studios. Limited to 18 seats per wave so reviewers can read every artifact.