Job Type: Versátil Contract (Remote)

Cobalt is hiring PhD-level biologists to review and build tasks that test whether frontier AI models can reason through real experimental science. Your job is to judge whether a model’s reasoning about an experiment is correct, and to explain precisely where it goes wrong when it is not. We are especially looking for scientists with hands-on flow cytometry experience.

If you are the person in your lab who spots the flawed control, the mislabeled gate, or the conclusion the data does not support, this role is built around that skill.

What you will do

- Review AI-generated answers to experimental biology problems end to end, including the experimental design, the methods, the data interpretation, and the final conclusion, and confirm that each step is scientifically sound and consistent with the others.
- Write realistic problems from your own area of expertise, along with reference solutions that show the reasoning a skilled scientist would follow, not only the final answer.
- Look for ways a model could reach a correct-looking answer without sound reasoning, such as relying on a textbook pattern that does not fit the data, ignoring a missing control, or overlooking an artifact, and document each issue clearly.
- Judge whether a model’s failure reflects a real gap in scientific reasoning, or instead reflects an ambiguous question, missing information, or an unfair reference answer.
- Write clear, specific feedback and work with task authors through revisions until each task meets the acceptance standard.
- Join occasional expert calls of 30 to 60 minutes with research teams who want to understand how experts approach a specific problem.

Example tasks you might review include designing a multicolor flow cytometry panel for a given set of markers and instrument,



critiquing a gating strategy and identifying where a population was misassigned, explaining why compensation errors produced a false double-positive population, and judging whether a reported difference between treatment groups is supported by the data.

Who we are looking for

- A PhD in biology, immunology, cell biology, biochemistry, or a closely related life sciences field.
- Hands-on laboratory experience, gained through research, a thesis, a postdoctoral position, a core facility, or industry work. You should be comfortable reading an experimental protocol and saying whether it would work.
- Strong working knowledge of experimental design, including controls, replicates, and when a difference between two results is meaningful.
- Careful, direct written communication. Much of the job is explaining precisely what is wrong and why.
- No prior experience with AI is required.

Strong advantages

- Hands-on flow cytometry experience, including panel design, compensation, gating, and data analysis, using instruments such as the BD FACSymphony or Cytek Aurora and software such as FlowJo, FCS Express, or OMIQ.
- Experience with spectral or mass cytometry, or with clinical flow cytometry such as leukemia and lymphoma immunophenotyping.
- Experience managing or working in a core facility, training other scientists, or serving as a peer reviewer for a journal.

Why Cobalt AI

- Shape how AI understands science. Your judgment directly determines which tasks are used to measure what leading models can and cannot do in biology.
- Use your expertise in a new way. You will see up close how frontier models reason about experiments and where they break.
- Fit the work around your current role.

Set your own schedule, with flexible weeks that can scale up or down around your research and other commitments.

- Grow with us. Strong reviewers can move into task design, project leadership, and research collaborations on future projects.

📌 Biology Subject Matter Expert (PhD) (España)
🏢 Cobalt
📍 España