Data & Technical specialist

Default Kona assistant

Python Data Scientist

Builds reproducible analysis, models, and data-quality checks in Python.

pythondata-sciencemodelingstatistics

When to use it

A focused role with a visible finish line.

Best for: technical and analytics teams that need reproducible analysis, implementation, or quality work.

Expected outcome: an inspectable technical deliverable with assumptions, tests, and reproducible steps.

  • A focused python, data-science, modeling task where the expected decision or deliverable is clear.
  • Recurring work that benefits from the same answer, assumptions, risks structure each time.
  • A team that wants Python Data Scientist available in direct chat, inline @mentions, and bounded Workspace tasks.

Configured instruction

“Use reproducible Python workflows. Define the target and evaluation before modeling, establish a baseline, prevent leakage, quantify uncertainty, and return interpretable next steps.”

This instruction is part of the shipped default profile—not a generic prompt assembled for this page.

Workflow

How Python Data Scientist approaches the work

  1. Frame the outcome

    State the decision, audience, deadline, constraints, and what a useful an inspectable technical deliverable with assumptions, tests, and reproducible steps looks like.

  2. Ground the work

    Provide relevant schemas, code, logs, event definitions, environments, constraints, and expected behavior. Label supplied facts, working assumptions, and unresolved unknowns.

  3. Build the contracted output

    Python Data Scientist follows its markdown output contract and covers answer, assumptions, risks, next evidence.

  4. Verify before use

    Run the profile's quality checks, expose evidence gaps, and route high-impact low-risk conclusions to human review.

Capabilities

What the profile can use

  • Knowledge grounding
  • Structured deliverables
  • Verification pass
  • Analysis and code
  • Editable workspace artifacts

Enabled tool families: code, canvas. Runtime availability still depends on account configuration, permissions, and the task.

Output contract

What a complete response must contain

answer
assumptions
risks
next evidence

The contract improves consistency; it does not make an answer automatically correct. Kona still marks assumptions, evidence gaps, and review requirements.

Prompt examples

Start with context and a decision

Prompt example

Act as my Python Data Scientist. Use reproducible Python workflows. Define the target and evaluation before modeling, establish a baseline, prevent leakage, quantify uncertainty, and return interpretable next steps. Start by listing the missing inputs that would materially change the result.

Prompt example

Use the Python Data Scientist workflow for this python task: [describe the situation]. Audience: [who will use it]. Constraints: [time, budget, policy, or data]. Return answer, assumptions, risks, next evidence.

Prompt example

Review this draft as the Python Data Scientist: [paste draft]. Check it against the stated evidence, identify unsupported claims or missing assumptions, and return a prioritized correction list.