AI Reasoning Analyst (Charts & Data)

Turn your chart analysis skills into work that trains frontier AI

If you can read a dense chart and reason your way to the answer, we want your help. You'll build sharp questions about real data visualizations, work out the correct answers, and catch where today's best AI models get them wrong on your own schedule, from anywhere in the US.

Workada connects skilled contributors with meaningful AI work. Your judgment makes AI systems smarter, safer, and more useful for everyone. This project is all about charts and graphs: the kind of multi-step visual reasoning that people do easily and models still stumble on.

What you'll do

Each Task is built around a single chart or graph. You’ll be responsible for creating a challenging question about the visualization that AI models are unable to solve. Your workflow will include 5 steps

• Start with a chart or diagram. Use one from your own field or coursework, or pick from a set we provide. The harder the chart, the better, dense legends, multiple axes, overlapping series, technical domains
• Ask a real question. Write a question that takes genuine reasoning to answer; tracing lines, counting points, comparing regions, reading across scales
• Build the right answer. Work out the correct answer and a clear step-by-step solution that shows exactly how you got there
• Test it against AI. Run your question and chart through the model and see how it reasons
• Find where it fails. Refine until the model gets it wrong, then capture the failure alongside your golden answer and solution

What makes a strong chart task:

• A complex visualization that rewards careful reading (multi-axis, overlapping data, a technical domain, etc.
• A question with one definitive, checkable answer; reached through several reasoning steps.
• A clean step-by-step solution a reviewer could follow and verify.
• A task that AI gets wrong.

Charts you can bring:

We'd love for you to bring charts from your own field, coursework, or work; anything you have the right to use. Almost any data visualization can become a Task, for example:

• Time-series and financial charts (price and market charts, technical indicators)
• Scatter, dot, and bubble plots; especially dense or multi-series
• Multi-axis, stacked, or grouped charts with detailed legends
• Maps and spatial data (choropleths, geographic or weather overlays)
• Domain-specific figures and diagrams (engineering, materials, scientific, electrical or architectural
• Dashboards and multi-panel infographics

Who we're looking for:

Sharp quantitative thinkers who love reading data closely; students, academics, and professionals alike.

You'll need:

• A background in a quantitative field: engineering, economics or finance, architecture, or STEM (biology, chemistry, physics, earth and materials science, and more) as a student, recent grad, academic, or professional.
• Strong quantitative and analytical reasoning: you can work through a multi-step problem to a definitive, defensible answer.
• Fluency reading complex visualizations: charts, graphs, diagrams, and maps with multiple axes, dense legends, or overlapping data.
• Clear written English to explain your reasoning step by step, the way a reviewer can follow.
• Precision and attention to detail; small misreads are exactly what we want to surface in the AI.
• Comfort using AI chat tools and iterating on prompts (no coding or ML background needed).
• Based in the US, with the self-direction to work on a flexible schedule.

Bonus points:

• Access to complex charts or figures from your own coursework, research, or work.
• Deep familiarity with a domain full of rich visualizations (financial markets, materials science, GIS/mapping, and the like).
• Experience with data analysis, technical writing, tutoring, peer review, or AI/data work.

What We Offer:

• Pay 3 times a week through Stripe.
• 100% remote, fully flexible. Pick up tasks around classes, research, or work. No cap on hours.
• Grow with it. Clear training, feedback that sharpens your skills, and a community of highly supportive contributors.

Ready to apply? Join a team helping build the data foundation behind better technology.