HireMatch

About this project

HireMatch is our double-sided marketplace connecting businesses with virtual assistants, using vector embeddings to score matches and an in-app helper bot to guide hiring.

The Challenge

Matching businesses to the right virtual assistant across skills, time zones, and working styles is hopelessly inefficient under manual filtering. Keyword search misses qualitative strengths entirely, which produces poor matches and high turnover.

The Solution

We engineered an automated matchmaking engine that generates high-dimensional vector embeddings of business requirements and assistant profiles using OpenAI embeddings. Matches are scored and ranked instantly in PostgreSQL. An in-app helper bot powered by Claude 4.5 assists businesses in refining their hiring criteria.

Semantic matchmaking engine parsing qualitative

Semantic matchmaking engine parsing qualitative job requirements.

Interactive VA marketplace directory with

Interactive VA marketplace directory with real-time video portfolios.

Integrated hiring pipeline with scheduler

Integrated hiring pipeline with scheduler and Stripe contract payments.

In-app AI assistant guiding employers

In-app AI assistant guiding employers through onboarding and hiring.

AI Intelligence

Matchmaking engine powered by OpenAI embeddings, alongside an in-app conversational helper bot running on Claude 4.5 for recruitment support.

System Architecture

Next.js 14 client communicating with an Express backend, querying a vector-enabled Supabase database.

Database & Caching

Supabase PostgreSQL with pgvector for embedding searches, cached with Redis for high-speed catalog searches.

Messaging

Push notifications for match recommendations, Daily.co integration for in-app video interviews, and SendGrid email notifications.

Frontend

Backend

Deployment

Frontend hosted on Vercel, backend servers deployed on Heroku with Supabase backend.