Estimate infrastructure or operational capacity needed for Learning at global scale

Problem Statement Description

Product context: LinkedIn is Microsoft's professional network; its products include profiles, feed, jobs, recruiting, LinkedIn Learning, sales tools, messaging, and ads.

LinkedIn Learning serves a global professional audience with video-based courses, skill assessments, recommendations, enterprise access, and usage analytics. In this guesstimate, you are asked to estimate the infrastructure or operational capacity required to support Learning at global scale, with particular attention to how recruiters and talent organizations may use learning signals, skill development, and workforce upskilling workflows.

Frame the problem as a capacity-estimation exercise, not as a product strategy recommendation. You should define what “capacity” means for your estimate: for example, video streaming and storage, daily/monthly active learners, concurrent usage, course catalog operations, recommendation/query volume, enterprise reporting, customer support, content review, or a combination of these. Be explicit about which capacity dimension you choose and why it matters for LinkedIn’s Learning ecosystem.

Your estimate should account for LinkedIn’s global network context: members across regions and devices, enterprise customers, recruiters looking for credible skill signals, and professionals consuming learning content at different frequencies. You may make reasonable assumptions, but they should be clearly stated, structured, and testable.

The experience should consider:

- The scope and unit of capacity being estimated, such as peak concurrent streams, monthly learning hours, storage needs, support volume, or content operations workload.

- The relevant population base, including LinkedIn members, active learners, enterprise seats, recruiters, or talent teams.

- Adoption and frequency assumptions, including how often users consume courses, complete modules, search for learning content, or view skill-related signals.

- Regional and temporal variation, such as workday peaks, time zones, mobile versus desktop usage, and enterprise learning cycles.

- Infrastructure drivers, including video bitrate, course length, caching/CDN needs, recommendation requests, analytics events, and reporting workloads.

- Operational drivers, such as content ingestion, localization, quality review, instructor support, customer success, or trust and safety processes.

- Sensitivity of the estimate to key assumptions, especially active-user penetration, average watch time, completion behavior, and peak-to-average usage ratios.

- Sanity checks against comparable large-scale professional, video, SaaS, or learning platforms without relying on exact proprietary data.

Your goal is to produce a clear, defensible estimation framework that an interviewer can follow end to end. The final number matters less than the quality of your scoping, assumptions, arithmetic structure, sensitivity analysis, and ability to connect the estimate back to LinkedIn Learning’s global professional and recruiter-facing use cases.