We are looking for people who want to work on the hardest problems in AI, and who care deeply about who that AI serves.
Join us
We are hiring researchers, engineers, and operators who want their work to matter at the scale of the world.
See open roles
Why AIC
At AICONSORTIUM, you are not a cog in a pre-built machine. You are helping design the machine itself. We are at an early stage, which means your contributions shape the architecture, the culture, and the direction of an organisation that intends to be among the most important in AI. We offer meaningful equity, strong salaries, and, more importantly, work that you will be proud to describe to your grandchildren.
Open Roles
Drive our federated MoE research programme. Design privacy-preserving training protocols and evaluation frameworks for distributed expert networks operating across sovereign compute boundaries.
Squeeze every last token per second out of our inference stack. Work on quantisation, speculative decoding, continuous batching, and custom CUDA kernels for our sovereign compute fleet.
Build the training data pipelines, tokenisation strategies, and evaluation benchmarks that will make Clark genuinely fluent across the world's major and low-resource languages, not just transliteratively competent.
Shape the training data and evaluation methodology for Sattam.ai. Requires deep familiarity with law across jurisdictions, spanning constitutional, civil, criminal, and regulatory practice, and an interest in how AI reasoning interacts with legal argument.
Build the product surfaces that bring Clark and Sattam.ai to users. Own the full stack from API design to frontend UX. You care about performance, accessibility, and shipping things people actually use.
Translate complex AI concepts into clear, honest documentation. Write model cards, API references, tutorials, and policy summaries for audiences ranging from developers to regulators to the general public.
Culture
We write things down. Decisions are documented. Research is shared internally before it is shared externally. We do not run on tribal knowledge.
We disagree and commit. Debate is vigorous and expected. Once a decision is made, the whole team executes. We do not relitigate closed questions.
Safety is everyone's job. Every engineer, researcher, and writer is trained on our safety principles. There is no safety team that handles safety so others do not have to.
We take rest seriously. Burnout produces bad science and bad products. Sustainable pace is not a perk. It is a research methodology.
Ready?
We welcome speculative applications from exceptional people. Tell us what you are working on and why you want to build with us.
