← All projects
Project — Product / Startup

Loom.

An anti-swipe campus matchmaking app that replaces infinite scrolling with two intentional daily matches powered by NLP-based compatibility signals.

The product bet

Most dating apps optimize for more swipes. Loom tests the opposite idea: fewer choices, stronger context, and a smaller set of daily matches.

From problem to product

What Loom does differently.

Problem
  • Infinite scrolling
  • Shallow profile judgment
  • Choice overload
  • Low trust
Loom direction
  • Two daily matches
  • Prompt-based profiles
  • Campus verification
  • NLP-based compatibility signals
How Loom works

From prompt to pair.

  1. 01
    Vibe Check

    Users answer prompts around intent, personality, and communication style.

  2. 02
    Relational Filters

    Baseline compatibility rules remove obvious mismatches.

  3. 03
    Text Embeddings

    Prompt responses are converted into compatibility signals.

  4. 04
    Daily Match Pair

    Users receive two intentional matches instead of an infinite feed.

What I owned

End-to-end ownership.

01

Product concept

Framed the anti-swipe thesis and defined Loom's core product principles.

02

Frontend MVP

Shipped a deployed web client with a Neon-Noir visual system and dual-card daily match flow.

03

Matching architecture

Designed the pipeline: relational filters, text embeddings, PostgreSQL, pgvector, cosine similarity.

04

Launch strategy

Planned closed campus launch with .edu verification to test match quality and retention early.

Technical architecture

System flow.

Frontend MVP
Prompt Inputs
Relational Filters
Text Embeddings
pgvector / Cosine Similarity
Daily Match Output
Current status

The deployed site is a frontend MVP and prototype interface. It demonstrates the daily match experience, visual system, and core product surfaces. Production database vector operations and live cross-user matching are not yet in production.

Supporting material

Artifacts.

Next project
Compass →