Location: Fully remote, with a preference for the US, Canada, or Europe
Start date: ASAP

Public AI is building a full-stack open-source competitor to ChatGPT and Claude, governed as public infrastructure. The stack we’re building isn’t one codebase. It’s dozens: open-source models from sovereign labs, open-source inference servers, frontends, evaluation frameworks, MCP servers, agent tooling. None of these were designed to work together. Our job is to make them compose well enough that a user, a government, or a downstream builder can treat them as a single product.

As Open Source Engineer, you’ll be the person building the connective tissue: writing the PRs that make components interoperate, building the demos that prove the stack is real, and filling the gaps that show up when you actually try to compose things. You’ll report directly to the CTO.

You will

This is a hands-on integrator role. In your first three months, you’ll likely:

  • Submit compatibility PRs across the open-source and open-weight components we depend on — OpenWebUI, vLLM, LiteLLM, evaluation frameworks, MCP servers, and the rest. Upstream where possible, in our forks where not. You’re comfortable opening a PR against a project you don’t maintain and shepherding it through someone else’s review process.
  • Build reference integrations and public demos that show how the pieces fit. When a partner asks what it actually looks like to run Apertus on sovereign HPC through chat.publicai.co with an open evaluation overlay, the answer should be a URL, a notebook, and a short video, not a roadmap.
  • Identify and fill gaps in specific OS projects where the stack is close but not quite there for real deployments. Some gaps you patch yourself. Some you scope crisply and hand to the right maintainer, sometimes with a Sovereign Tech Fund grant attached.
  • Ship the unglamorous integration work that makes new contributions land cleanly: templates, contribution guides, CI scaffolding, the infrastructure that determines whether a partner’s donation becomes a working integration or sits in a fork for six months.
  • Produce the public artefacts of the stack working together — tutorials, cookbooks, “build a sovereign chat assistant” walkthroughs. We have a monthly drumbeat of releases to sustain, and a lot of it needs to be technical material.
  • Track the open-source AI stack as it ships. OpenWebUI ships fast. vLLM ships fast. New things appear weekly. You’ll keep us current.

You’ll have wide latitude. We need someone who sees what’s broken between two systems and fixes it, not someone who waits for tickets.

What we’re looking for

Required

  • Real open-source contribution history, especially merged PRs to projects you don’t own. Bridging two ecosystems is the strongest signal.
  • Comfort across the modern open-source AI stack, e.g. OpenWebUI, vLLM, LiteLLM, Hugging Face tooling, MCP, or similar. You don’t need to have shipped all of them. You do need to be the kind of person who can read an unfamiliar codebase on Tuesday and file a useful PR by Friday.
  • Strong Python and enough TypeScript to be useful at the frontend boundary. Systems literacy to debug across the inference / orchestration / UI layers.
  • A taste for integration work. Some engineers light up building new things. We need someone who lights up making two existing things talk to each other.
  • Writing ability. PR descriptions, issue reports, tutorials, RFCs to partner teams. A lot of this job is writing.
  • Ability to work mostly autonomously in a small team and travel occasionally for team workshops.

Most important: intrinsic motivation.

This is a nonprofit, open-source project building public goods. We need someone who actually cares about that mission and wants to build infrastructure for the public, not for a series B. We’re looking for smart, creative, scrappy, and moderately rebellious people who will challenge themselves, challenge us, and push for better. If you need a manager to tell you what to ship next, this isn’t the right role.

Nice to have

  • Prior experience in a large-scale collaborative open-source AI effort (BigScience, EleutherAI, OLMo, Apertus, EvalEval, similar)
  • Familiarity with sovereign or HPC compute environments (CSCS, LUMI, EuroHPC, Barcelona Supercomputing Center)
  • Experience with evaluation infrastructure (HELM, LM Eval Harness, HAL, Docent, ARIA)
  • Multilingual capability, especially in Apertus target languages
  • A track record of turning messy multi-repo systems into coherent platforms

Logistics

  • Location: Fully remote, with a preference for the US, Canada, or Europe. Must be available for occasional team workshops.
  • Employment: Starting as a full-time contract, with the option to graduate to a full-time employee role.
  • Compensation: Starting at $50,000 USD over three months, with room to scope upward for exceptional experience.
  • Reports to: CTO

Send a note to josh@publicai.co with whatever best represents you — CV, GitHub, projects you’re proud of, integrations you’ve stitched together, things you’ve broken and fixed. Tell us why public AI matters to you.

We read every application.

← Back to all jobs