Ten Rupees a Ride

Indian Railways moves roughly 23 million passengers a day. A long-distance journey runs for hours. A suburban commute runs twice a day, every day, for years.

Mobile signal along a railway line is patchy and drops out entirely for long stretches. So passengers either download something before they board, or stare out of the window.

Nothing is broken. This is an opportunity that nobody has taken at a price ordinary passengers would pay.

The target set for you is ten rupees per person per ride. That number is the whole problem. It rules out almost every business model that works elsewhere, and it is low enough that a passenger might actually pay it without thinking.

Two reasons, and only one of them is technical.

The technical one is that you cannot stream from the internet on a moving train, so content has to already be on board. Then one server has to serve hundreds of phones at once over wireless, in a metal tube, while everyone is watching something different. Wireless capacity, not storage, is what breaks first.

The commercial one is harder. At ten rupees a head you cannot license premium film content. The rights alone would eat the fare several times over. Indian Railways has announced its own on-demand video, news and multilingual content on trains, and edge-cached video for rail operators is already a commercial product internationally, so you would not be first.

• Branch — Electronics for the wireless and hardware side, CS and IT for the software and the caching. One of the more balanced problems in this pool.
• Technical skills — Wireless networking and capacity planning, local caching and content delivery, adaptive streaming, embedded and low-power hardware, load testing at realistic scale.
• Other skills — Unit economics, content licensing basics, reading a regulated environment, testing a claim rather than demonstrating a feature.