Skip to content
Magrus
Menu

Case 07 / 08

TwinlyAI

Resumes become AI twins recruiters can interview, backed by RAG search.

TwinlyAI homepage: 'Your resume, now alive.' above a preview of the product.
Year
2025–26
Type
AI SaaS
Disciplines
Applied AI, Full stack engineering, Product design
Stack
Next.js, React, Tailwind CSS, FastAPI, MongoDB, Qdrant vector database, AWS

02 · Objective

What it had to do.

A resume reduces a person to keywords, and recruiters have too many to read. Candidates should get to show how they think.

03 · Approach

How we went about it.

  1. Resume parsed and split

  2. Indexed in a vector database

  3. Answers only from what's retrieved

  4. Candidates can connect GitHub

04 · Features

What we built.

A twin grounded in the resume
Recruiters ask; the candidate's own material answers.
GitHub sync
Repositories feed the twin, so it discusses real projects.
Search by meaning
Find candidates by what they've done, not keywords.
API access
Connect twins to your own hiring tools.

05 · System

How the parts connect.

TwinlyAI systemParse & embed6Vectordatabase7Candidate1AI twin chat8Resume upload2Recruiter4Semanticsearch9GitHub sync3API keys5
Layers
  • 01Build
  • 02Intelligence
Parts
  • Surface: What people touch
  • Engine: The logic and AI that does the work
  • Store: Where data lives
  • Channel: How people and data get in and out
  • Person: Who approves
Lines
Data moves between these parts
  • Build layer
    • Candidate
    • Resume upload
    • GitHub sync
    • Recruiter
    • API keys
  • Intelligence layer
    • Parse & embed
    • Vector database
    • AI twin chat
    • Semantic search
  • Connections
    • Candidate connects to Resume upload
    • Candidate connects to GitHub sync
    • Resume upload connects to Parse & embed
    • GitHub sync connects to Parse & embed
    • Parse & embed connects to Vector database
    • Vector database connects to AI twin chat
    • Recruiter connects to AI twin chat
    • Recruiter connects to Semantic search
    • Semantic search connects to Vector database
    • API keys connects to AI twin chat

06 · Craft & performance

The details underneath.

  • Streamed responses
  • Three sign in providers
  • Walkthrough video sized to the screen
  • Reduced motion respected

07 · Results

Where it stands.

In private development. Usage shared once it's measured.

Credits: the team.

Want a system built like this?

Start a project

Answer a few short questions. We reply within one working day.