The setting
is an open-source backend platform built on Postgres. Its runs as a hosted service that Supabase manages, or as packaged images that customers host themselves.
Planning another batch of Spark projects puts spending certainty against uninterrupted experiments.
A firm allowance protects a developer's budget, while an estimate leaves room for workloads whose lifetime is still unknown.
Who you’d be doing this for
“My founder asked what another hundred experiments would cost. I spent the afternoon guessing how long each one would stay awake.”
Javad Tehrani · Backend developer at a small application platform · a character in this brief
Runs batches of experimental databases and explains their spending to the engineering founder.
What is at stake
Eleven of 15 developer teams estimate database spending outside the product. You have to choose what certainty the first version offers without turning a forecast into a promise or a budget into an outage.
Why it isn’t already fixed
Every obvious fix costs something else. That’s the part you’d have to decide.
- budget certainty vs uninterrupted experiments
- useful estimates vs honest uncertainty
- narrow charge coverage vs a complete spending decision
- planning flexibility vs prepaid predictability
Why Supabase
Spark projects provision in a second and scale down when idle, so project count alone does not tell developers what their experiments will cost.
Written with these in mind
Not your kind of problem? 27 more at Supabase, or browse every organization.
Where the organization context comes fromThe postings this brief was written from, and the reporting read before it.
The organization context comes from public information. The situation and characters are written for the brief.
Job postings it was written from
Public descriptions of the work at Supabase. They are listed as sources, not as openings. The lines under a posting are its own words, where it meets this brief.
- Product Manager - Administration & Governance (opens in a new tab)
- Define what success looks like, instrument it end to end, and let the data guide what gets built next—from time to first project and self-service adoption to production usage and expansion across the company.
- Know your way around controlled experimentation and data analysis, and can connect easier adoption to sustained product use.
- Product Manager - Spark (opens in a new tab)
- Spark projects provision in a second, scale down when idle, and are cheap to enough to use liberally, guilt-free. while keeping the Supabase interface agents already know.
- This role owns establishing Spark as the default starting point for agent-built software and how projects grow from an early experiments into production workloads on Supabase.
- Product Manager - Strategic Partner Integrations (regular travel to Bay Area) (opens in a new tab)
- For a platform built on Postgres with Auth, Compute, Storage and generated Data APIs at its core, the surfaces where Supabase meets Partners and their systems are some of the most important avenues that we own - this role owns these surfaces.
- Build with what you own, using the surfaces on your roadmap the way a Partner would: standing up an auth flow, calling the Data API, deploying a database, wiring Supabase into another product.
- Product Lead - Infrastructure (opens in a new tab)
- You talk to customers across the full spectrum: developers running one small project, startups whose bill is mostly compute and disk, enterprise architects running a network or security review, and partners building on Supabase.
- Our customers range from a developer on a free project to a regulated enterprise that routes traffic through PrivateLink https://supabase.com/docs/guides/platform/privatelink and puts us through a security review before it signs.
Reporting read before this brief was written
The pieces closest to the situation in this brief.
On September 30, 2026, AWS announced that it had embedded DuckDB in Amazon Aurora PostgreSQL (17.11+ and 18.6+), allowing queries over Iceberg tables and Parquet files in data lakes without ETL, at no extra licensing fee.
AWS Embeds DuckDB in Aurora PostgreSQL, Enabling Direct Data Lake Queries Without ETL (opens in a new tab) · finance.biggo.com · Oct 1, 2026
This is the setup. The work is inside.
Running it puts you in the room: the full situation and its constraints, stakeholders who push back in their own words, and the decisions that are yours to make. What you produce becomes a Day One Plan — work you can show someone instead of describing.
Where similar work appears
Other live briefs connected to this one, and what connects them.
Similar problems at other organizations
Like this one, each of these turns on deciding on evidence that's still thin.
Data engineers can't tie costly Apache Spark workloads to an owner or a usage-control action until spend has built up.
You have to weigh fast, useful attribution against calculations users can trust across clouds.
Administrators are rebuilding AI usage forecasts from exports because no direct limit exists today.
You must weigh firm spending boundaries against an admin control that developers can understand and work within.
Pilot customers cannot tell whether their next 30 days of GPU demand fits their current quota.
You must weigh a useful planning signal against the risk of implying guaranteed capacity.
The same setting, from another role
Each is set around Supabase’s Supabase Postgres Database, written for a different role. They’re separate briefs, not the same scenario.
Engineers are reopening deployment alerts because the evidence needed to judge severity is scattered.
You must decide what AI gathers, what it may suggest, and where an engineer must review it.
Repeated Nix packaging failures are holding PostgreSQL releases while engineers reread the same build evidence.
You must decide what the system can verify itself and where an engineer must take over.
Recovery investigations repeatedly produce competing explanations from the same logs and failed tests.
You must decide what evidence the system can assemble and what an engineer must still confirm.