@apriliyaay_: 🤚#fyp #madiun24jam

liyass_
liyass_
Open In TikTok:
Region: ID
Sunday 27 September 2026 17:52:01 GMT
561
42
2
3

Music

Download

Comments

amildatrihapsari
amelllaaa🇮🇩🇭🇰 :
jannnn kangen pollll
2026-10-01 14:57:08
0
hass6_
𝙃𝙖𝙨𝙨𝙆𝙖𝙜𝙚𝙣𝙤𝙪 :
lb
2026-09-28 03:07:10
0
To see more videos from user @apriliyaay_, please go to the Tikwm homepage.

Other Videos

Tinder's hardest problem isn't the swipe animation. It's deciding which one person, out of millions, you see next — and making that decision feel personal in under half a second. Here's what actually happens between you opening the app and one card appearing: 1. Your phone sends a location. The backend does NOT measure the distance to every user — that's millions of calculations for one screen. It turns your coordinates into a geohash cell, searches that cell plus its 8 neighbours, and gets a shortlist. 2. Then the filters run, in order: 20 km, age range, gender preference, not blocked, not already swiped, active recently. Millions collapse to a couple of thousand. 3. Now the real work. Every survivor gets scored — distance, activity, preferences, past swipe behaviour, profile completeness, and the heaviest weight of all: the model's guess at whether you two would like each other. The deck IS that ranked list. 4. You swipe right. That's a tiny event, a few dozen bytes. It gets written, and then one index lookup asks the only question that matters: did they already swipe right on you? If yes, both phones have to learn it at the same instant. One write, or none — never one phone celebrating while the other has no record. Five systems, agreeing in a fraction of a second: geospatial · recommendation · matching · messaging · trust & safety Tinder does about 1.6 billion swipes a day. The swipe is the easy part. The order of the deck is the product. Which of the five would you build first? #systemdesign #softwareengineering #productdesign #techexplained #devcommunity
Tinder's hardest problem isn't the swipe animation. It's deciding which one person, out of millions, you see next — and making that decision feel personal in under half a second. Here's what actually happens between you opening the app and one card appearing: 1. Your phone sends a location. The backend does NOT measure the distance to every user — that's millions of calculations for one screen. It turns your coordinates into a geohash cell, searches that cell plus its 8 neighbours, and gets a shortlist. 2. Then the filters run, in order: 20 km, age range, gender preference, not blocked, not already swiped, active recently. Millions collapse to a couple of thousand. 3. Now the real work. Every survivor gets scored — distance, activity, preferences, past swipe behaviour, profile completeness, and the heaviest weight of all: the model's guess at whether you two would like each other. The deck IS that ranked list. 4. You swipe right. That's a tiny event, a few dozen bytes. It gets written, and then one index lookup asks the only question that matters: did they already swipe right on you? If yes, both phones have to learn it at the same instant. One write, or none — never one phone celebrating while the other has no record. Five systems, agreeing in a fraction of a second: geospatial · recommendation · matching · messaging · trust & safety Tinder does about 1.6 billion swipes a day. The swipe is the easy part. The order of the deck is the product. Which of the five would you build first? #systemdesign #softwareengineering #productdesign #techexplained #devcommunity

About