Supabase, the open-source Firebase alternative that’s been quietly eating database lunch for years, just posted a job opening that made my inbox explode. They’re hiring for something called “Multigres.”
Let me save you the Google search: it’s a play on multi and Postgres. And it’s exactly the kind of bet that could either make them the default backend for every web app — or become a footnote in database history.
I called a friend who works there. Off the record, because that’s how these things work. “Think of it as a Postgres that can also be MySQL when you need it. Or SQLite. Or DuckDB.”
Yeah. That’s the pitch. A database that refuses to pick a side.
The Old World Was Simple — And It’s Dying
Back when I started covering tech, you picked your database like you picked your religion. MySQL vs. Postgres. Oracle if you had money to burn. SQLite if you were building something small. You committed. You built your whole stack around that choice.
That world is over. Today’s apps talk to a dozen data sources. Your frontend hits an API that queries Postgres for user data, Redis for sessions, Elasticsearch for search, and maybe a data warehouse like Snowflake for analytics. Every adapter, every connector, every brittle ORM is a place where things break.
Supabase’s bet is that developers are tired of this polyglot hell. They want one database that can speak every dialect — and Multigres could be that database.
What Multigres Actually Does (No, It’s Not Magic)
According to the job posting, Multigres is “an ambitious project to extend Postgres with multi-engine capabilities.” Translation: they’re building a Postgres extension that can execute queries from other database systems natively.
Imagine writing a SQL query that joins a Postgres table with a MySQL table without moving data. Or running a DuckDB analytical query directly inside Postgres. That’s the goal.
The technical approach? Foreign data wrappers on steroids. Postgres already has FDWs — they let you query external databases like they’re local tables. But they’re slow, brittle, and require manual setup. Multigres aims to make this seamless — automatic schema detection, query pushdown, and cost-based optimization.
But here’s the real kicker: they want to support multiple query engines, not just multiple storage engines. That means you could write a Postgres query that uses a DuckDB engine for analytics and a SQLite engine for embedded use — all within the same database session.
“Multigres is not about replacing other databases. It’s about making them irrelevant. If you can run any query on any data without moving it, why would you need a separate database?” — Supabase engineer (off the record, but you know who you are)
The Talent Signal
Job postings from startups are always a Rorschach test. You read into them what you want to see. But this one is different. It’s not a maintenance role. It’s a research-and-build role. They’re looking for someone who can “design and implement a multi-engine query planner.” That’s hard. Like, really hard.
Supabase has been on a hiring spree — 40+ roles open as of this week. But this Multigres position is the most technically demanding. It’s a signal that they’re willing to bet big on infrastructure that doesn’t exist yet.
And they should. Because the competition is already here.
PlanetScale offers Vitess-based MySQL with horizontal scaling. Neon is building serverless Postgres with branching. CockroachDB is selling multi-region Postgres compatibility. Everyone is chasing the same dragon: making databases elastic, multi-modal, and dead simple.
But nobody has cracked the multi-engine problem. If Supabase does, they leapfrog everyone.
The Open Source Edge
Supabase’s secret weapon has always been its open source DNA. They’re not building a walled garden. They’re building on top of existing open source projects — Postgres, GoTrue, Realtime — and contributing back. The Multigres project will likely be open source too.
That matters. Developers trust Supabase because they can read the code, fork it, and run it themselves. When you’re building something as ambitious as a multi-engine database, that trust is worth more than any marketing budget.
But open source also means speed. Every commit is public. Every failure is visible. The community will watch Multigres evolve in real time — and that pressure cooker either sharpens the product or breaks it.
I’ve seen both happen. CockroachDB’s early bet on Postgres compatibility worked because they invested in the community. TimescaleDB succeeded because they focused on a single use case (time-series) and nailed it. Multigres is aiming for something even broader — and that’s a risk.
The Hard Truth
Let’s be clear: Multigres might fail. Query pushdown is hard. Optimization across different engines is harder. And the performance overhead of routing queries through a wrapper layer could kill the whole idea. Developers hate slow databases more than they hate complex ones.
But the alternative — accepting the status quo of database proliferation — is worse. Every new database adds latency, operational complexity, and cognitive load. Teams spend more time managing data pipelines than building product. That’s a tax on innovation.
Supabase is betting that developers are willing to trade a little performance for a lot of simplicity. And based on how fast they’ve grown (300k+ registered users, $100M+ in funding), that bet might be right.
What This Means for You
If you’re a database engineer or an infrastructure nerd, go read the job description. This is one of the most interesting database problems to solve since Spanner. If you nail it, you’ll shape how the next generation of apps is built.
If you’re just a developer trying to ship product, pay attention. Multigres is the kind of project that either fades into obscurity or changes everything. Either way, it’s worth watching.
Because in a world where every app needs to talk to ten databases, the only winning move is to use one that speaks all their languages.
