Preparing your design workspace

Loading the problem and your latest saved attempt…

Design a Student Analytics Dashboard

Design private student and course dashboards with durable result corrections, clear data freshness and a practical delivery plan.

Problem statement

For this exercise, assume a university wants a web application for student performance analysis. Students view their own assessment results and progress; instructors view only their assigned courses and can record or correct assessment results. A correction must not silently overwrite another accepted correction.

Course averages and completion counts must come from one completed refresh. Show both the data cutoff represented by that snapshot (asOf) and when calculation finished (completedAt); mark it stale when the represented-data cutoff is more than five minutes old, even if the refresh just finished.

The team already operates an authenticated API service, a relational database and scheduled workers. Reuse that platform where it fits; the exercise does not prescribe a programming language or vendor.

Propose the minimum first release, sketch its architecture, and explain how you would divide UI, API, data, testing, release and maintenance work. Give a justified staffing estimate for an eight-week first release. Predictive models, admissions, payments and cross-university data sharing are outside scope.

What to build

  1. Students retrieve their own results and progress; instructors retrieve summaries only for assigned courses.
  2. Authorized instructors record and correct assessment results, receive durable acknowledgement and detect conflicting concurrent corrections.
  3. Dashboards show course averages and completion counts from one completed refresh, with its represented-data cutoff (asOf), calculation completion time and an explicit stale indication when the cutoff is more than five minutes old.

At scale

  • For this exercise, assume one university with 20,000 students, 500 instructors and about 3 million assessment-result rows per term.
  • At peak, serve 200 dashboard reads per second and 20 result updates per second. Target read latency below 500 ms at p95 in a healthy single-region deployment.
  • A committed result change should appear in its course summary within 5 minutes. A failed refresh must not replace a complete summary with partial results.