First community: Thapar · @thapar.eduNot a feed · not a job board · not dating

Cynergi: say what you're building, meet who can help.

Say what you're building. Meet who can help. Describe your goal in a sentence. We find the few people worth talking to, and tell you why.

AKRS Example matchBuild together · reason shown to both
Find your peopleHow it works
BuildLearnExplore
Intros per request a few, explained Servers Singapore Local time IST --:--:--
Principle 01Intent first, skills secondOne engineSix intents

Intent before skills.

Most apps ask what you can do. We start with what you're trying to do: build something, swap a skill, find a mentor, stay on track. Then we look at skills.

Two spheres, one ink and one green, resting together on a white plinth.
6ways to ask, one matching engine, weighted differently for each.

Principle 02

Describe, don't fill forms.

Your wordsExample

Third-year CSE. I can teach Python and I want to learn product design. Building a budgeting app on weekends and I need a designer.

Offers
PythonTeaching
Seeks
Product designA designer
Intent
Build togetherSkill exchange
When
Weekends

You check what we understood and fix anything we got wrong. Your own words are always kept.

A field of profile cards seen from above, a few of them green.

Write a few sentences about who you are and what you want to do. No skill pickers, no endless dropdowns. A model reads it once and turns it into something the matcher can use.

How matching works

Every request starts with understanding what you're asking for. Then four steps narrow everyone down to a few.

Stacked discs on a dark floor, one layer in mint. A dark field of spheres with five white ones inside a mint ring. A descending row of columns, the tallest in mint. Two spheres on pedestals joined by a mint bridge.

You write a few sentences. A model reads them and pulls out what you offer, what you're looking for and what you're into. You review the result and correct it. Your own words are always kept.

Model reads your text once

Four parts of every profile become vectors: who you are, what you offer, what you seek and what you're into. A vector search and a keyword search narrow everyone to a few hundred people, after blocks and availability are applied.

4 facets · 384 dimensions

Candidates are scored on fit and on whether the match works both ways. Small penalties stop the same people being shown to everyone, but they never lift a poor fit over a good one. Anyone below the quality floor is dropped.

Weights applied in code, not by the model

A model reads up to fifteen candidates and scores how well each meets your need and how much you share. Anyone with a conflict is dropped. The rest come with a reason both of you see. If nobody fits, we say so.

No names or emails sent to the model
/0points for how well someone meets your need
/0points for what you share
/0matches is an allowed answer: "Nobody fits yet."

Every match here starts with a reason you can read.

Why you two · example

You need a designer for a budgeting app. They've built two Figma prototypes, want a real project to work on, and are free on weekends, same as you.

AKRS Shown to both people, before either says yes

Before you start.

Your name and links are shown only to people you are connected with. When you're suggested to someone, or send them an intro, they see your summary, skills, interests and availability, and why you were matched, but not your name or links.

No. We don't use AI to read your chat messages. Only you and the person you're talking to can read your chat.

Neither of you can message the other, and you won't be suggested to each other. They aren't told.

Cynergi starts with Thapar. Sign in with your @thapar.edu address. You need to be 18 or over.