Launching today

AetherGrid

Launching today

Open-source AI workload orchestrator, lease-based execution

1 follower

Open-source AI workload orchestrator, lease-based execution

1 follower

Visit website

AetherGrid is an open-source AI workload orchestrator built to survive real failure, not demo well. Jobs run under time-bound, continuously-renewed leases, not a simple assignment flag, so a crashed worker can't cause duplicate execution. Real subprocess execution with enforced timeouts, reconciliation that reclaims abandoned work, and 280 tests proving it, including one that force-kills a process ignoring SIGTERM. Clean Architecture, Domain-Driven Design,26 ADRs, live with CI/CD on every merge.

Free

Launch tags:Developer Tools•Artificial Intelligence•GitHub

Launch Team

Tines The single, secure environment for agents, apps, and automations.

Promoted

Maker

📌

Hey everyone, I'm Wycliffe, the maker behind AetherGrid. I got into this because of a very specific kind of pain. I've spent years tracing race conditions in production systems, first on revenue systems handling real money at national scale, then in worker queues at Toyota, and now reviewing LLM-generated backend code for a living. The pattern is always the same: the code works fine on the happy path, then falls apart the moment something fails mid-execution. A worker crashes, a network blips, a heartbeat drops, and suddenly two processes think they own the same job. Most scheduler side projects don't deal with that because they're a script polling a dictionary in a while loop. It works until it doesn't. I wanted to build the version that doesn't fall over, so job execution ownership in AetherGrid runs on time-bound leases that renew continuously for the job's whole runtime, not a boolean flag that lies to you the moment something goes wrong. The approach shifted a lot as I built it. I started just wanting a working scheduler, but the more I dug into failure recovery, the more I realized the interesting engineering problem wasn't scheduling jobs, it was un-scheduling them safely when things break. That's where leases, reconciliation, and the bounded retry logic came from. I ended up writing a test that genuinely kills a subprocess that ignores SIGTERM just to prove the timeout enforcement actually works under pressure, not just in theory. It's live in production now with 280 tests behind it and 26 ADRs documenting why every non-obvious decision was made. Happy to answer anything about the architecture, the lease mechanics, or why I went with raw psycopg over an ORM. Would genuinely love to hear how others have handled execution ownership in their own systems too.

Report

18h ago

Forum Threads

p/aethergrid

•

17h ago

Building AetherGrid: an AI orchestrator that survives real worker failures

I've spent years chasing race conditions in production systems, revenue systems handling real money, Kafka worker queues, and now reviewing LLM-generated backend code for a living. The pattern is always the same: code works fine until something fails mid-execution, then two processes think they own the same job.

AetherGrid is my answer to that. Job execution ownership runs on time-bound leases that renew continuously for the job's whole runtime, not a boolean flag that lies to you the moment a worker crashes or a heartbeat drops. Real subprocess execution with enforced timeouts, reconciliation that reclaims abandoned work, 280 tests behind it, including one that genuinely kills a process that ignores SIGTERM.

Live now on Product Hunt, launching August 21st. Would love feedback from anyone who's dealt with execution ownership or lease-based systems in their own infra, and happy to answer anything about the architecture.

Repo: https://github.com/wycliffRotich...

View all