5,322 active roles, straight from company career pages.
Data and AI work has a property that most disciplines envy: it is naturally async. A pipeline does not care what hour you trigger it. A model checkpoint does not need you in the same room as your reviewer. That structural reality is a big part of why remote data and AI roles have become a genuine career track rather than a pandemic-era workaround. JobFerret currently lists 5,322 active remote Data & AI jobs, every one of them remote by definition, because that is the only kind we index on this page.
What makes remote data teams function well is not just a good VPN or a shared Slack channel. It is tooling that travels. Version-controlled notebooks, reproducible environments, orchestration layers that log everything so a colleague picking up your work eight hours later can reconstruct exactly what you ran and why. The roles clustered here tend to assume that fluency. You will see job descriptions that mention dbt, Airflow, MLflow, or Weights & Biases not as nice-to-haves but as the shared vocabulary of the team. That is the culture of async data collaboration, and it shapes what hiring managers write and what candidates need to demonstrate.
Among the companies actively hiring on JobFerret right now, bjakcareer leads with 446 open roles, followed by agency (191), toloka-ai (141), futuresight (110), lilt-production (85), and toloka-annotators (60). The presence of annotation-focused employers like toloka-ai and toloka-annotators is telling: the human-in-the-loop layer of AI development is itself a distributed, async discipline, and it draws heavily from a globally dispersed talent pool.
Among roles with a stated location, the picture is deliberately global. The largest single location group is simply "Remote" (350 listings), followed by "World Wide - Remote" (184), United States (145), San Francisco (119), and London (94). That spread reflects something real about how data infrastructure teams are built: a data engineer in Warsaw can own the same pipeline as a machine learning engineer in Toronto, and neither needs to relocate. The 567 listings with no location specified at all reinforce the point: for a meaningful share of these employers, geography is genuinely not part of the conversation.
If you are moving into this space or deepening your position in it, the async-first mindset is worth treating as a skill in its own right. Documenting decisions in pull requests, writing runbooks that assume no institutional memory, building dashboards that answer questions before they are asked: these habits show up in job descriptions and in interviews, and they separate candidates who have worked in truly distributed data teams from those who have only worked remotely in name.
Requirements vary by employer, but many listings on this page assume familiarity with at least one orchestration or transformation tool. Reading individual job descriptions carefully will tell you which stack a team has standardized on, which matters more in async environments where onboarding happens largely through documentation.
Yes. The listings here span the full data and AI workflow, from labeling and quality review through to infrastructure engineering and applied research. Employers like toloka-ai and toloka-annotators represent the human-feedback side of the pipeline, which is increasingly central to how production AI systems are built and maintained.
JobFerret indexes new postings continuously, so the count of 5,322 active roles reflects the current state of the index rather than a periodic batch. Checking back regularly or setting up alerts is the best way to catch roles as they appear.