BaseThesis Labs, A frontier lab focused on democratising the way humans interact with technology to maximise their potential.
Hiring exceptional engineers to build with us.
What BaseThesis expects from you?
High ownership on what you ship and the results you drive.
Measurable results or clean negative findings. Clarity over everything else.
Exceptional engineering skills and managing production grade systems at scale.
BaseThesis affiliation on preprints, arXiv submissions, and conference papers
Open source release of models, code, and datasets on BaseThesis's GitHub and Hugging Face, unless explicitly scoped otherwise
Willingness to ship, as a paper, as a prototype that graduates to a product team, or as infrastructure other researchers use.
What BaseThesis provides you:
Dedicated compute for your program = no GPU rationing, no shared queue waits.
A principal investigator on a 24 month horizon, resourced to produce one finding the field did not believe was possible.
Primary authorship on your research papers, with editorial feedback on drafts.
Reimbursements (travel, stay, attendance) to present your work at top venues = main tracks only at NeurIPS, ICML, ICLR, CoRL, RLC.
Access to our internal community of founders, researchers, systems engineers, and product builders working across the lab.
Unconventional Applied Research; Seeing your research deployed inside real businesses through BaseThesis's portfolio companies, not stopping at the preprint.
Co-authorship on cross-program collaborations, proportional to contribution.
Distribution of your work through BaseThesis's channels when it ships.
Role:
Applied AI Engineer / Systems Engineer / Full-Stack Engineer
Ship, maintain and improve production systems end-to-end, owning the path from research prototype to deployed infrastructure.
Distributed training or real-time inference at scale. Eg., multi-node orchestration, low-latency serving, compiler-level optimization where needed.
Fluent across the entire stack from model code, data pipelines, APIs, to frontend when the system needs one. The boundary between "ML" and "engineering" should feel arbitrary to you.
Multi-cloud orchestration, petabyte-scale data pipelines, or custom silicon integration, depth in at least one, literacy in the rest.