Small team, hard problem.
widemem is a small, remote team building memory infrastructure for LLM agents. Open source, local-first, Apache 2.0. We care about correctness, honest benchmarks, and code you can read before you trust it.
We are not actively hiring for a specific title today. That is the honest answer, and we would rather say it than post roles we are not ready to fill.
But if you are exceptional and this problem keeps you up at night, memory for agents, retrieval that holds up under scrutiny, systems that stay correct in regulated environments, reach out anyway. Send what you have built, in code or in prose, and tell us what you would want to work on. We read every message from someone who clearly cares.
How we work
Remote, async, small. Ship in public, correct in public. No feature theater, no vanity benchmarks. If a number is wrong, we fix it where everyone can see.
Who fits
People who care more about the thing being right than about being seen to be right. Deep on retrieval, systems, or applied ML, and allergic to hand-waving.