1 rider. 50 drivers nearby. Who gets the ride?
I drew out Uber's design. The hardest part: each driver offered a ride is locked in Redis with a 10s TTL. Accept → it's theirs. Ignore → lock expires, next driver.
No driver gets two rides. No ride gets two drivers.
3/ Catch: GEO members never expire. A driver whose app dies stays in the set forever. Fix: a heartbeat key with a TTL that matching checks first.
4/ Full design by Hello Interview: hellointerview.com/learn/sys…
A person: I earned ₹1 lakh.
A growth person:
Coming from a tier-1000 college nobody has ever heard of, my laptop was dead, my charger was dead. I literally ran my laptop on petrol, which I bought with udhaar ke paise.
And somehow, I turned that situation into my first ₹1 lakh.
Just opened my own Pahadi shop.
If anyone wants to buy something, feel free to contact me.
AI can’t replace me now.
2 million people. 1 seat. Who gets it?
That's the real problem behind Ticketmaster. I studied the design and drew it out 🧵
→ Seat lock, the hardest part:
Pick a seat and it's locked in Redis with a 10-min TTL.
Pay in time → it's yours.
Don't → the lock expires and the seat is free again. No cleanup job needed.
Everything else in the design exists to protect that one lock.
Full design by Hello Interview: hellointerview.com/learn/sys…










