Erwin's Hire AI
An AI hiring copilot, built end to end in 72 hours.
- Years
- 2024
- Type
- Personal
- Discipline
- Web Dev, Full Stack, AI
- Stack
- Next.js, React, Typescript, AWS
Erwin's Hire AI is an AI-powered hiring copilot designed to automate and compress the recruitment pipeline - from resume ingestion to interview evaluation - into a single, cohesive system. The project was conceived, built, and deployed within a 72-hour build window, with a deliberate focus on execution speed, system design, and end-to-end flow.
Recruiters can upload resumes which are automatically analyzed, indexed, and stored in a private datastore, ensuring complete data isolation per recruiter. Candidate discovery is enabled through text and voice-based search, combining semantic understanding with structured signals such as skills, projects, and experience to surface relevant profiles.
Shortlisted candidates are added to hiring campaigns and automatically invited via email using unique interview links. Interviews are conducted by an AI avatar with voice interaction, capturing timestamped responses to support integrity checks and detailed evaluation. Post-interview, AI-driven analysis ranks candidates and can trigger automated follow-up or confirmation emails based on recruiter-defined criteria.
The system was built using Next.js without a dedicated backend to maximize iteration speed, orchestrating multiple AI services - including TTS/STT pipelines and concurrent LLM personas - and deployed as a fully functional demo on Vercel, supported by a walkthrough video.
What I did
- 01Built an end-to-end AI hiring copilot within a 72-hour time constraint
- 02Implemented secure resume ingestion, private indexing, and candidate isolation per recruiter
- 03Enabled voice and text-based candidate search using semantic and contextual matching
- 04Designed an AI-driven interview system using a talking avatar with real-time voice interaction
- 05Orchestrated multiple specialized AI personas for concurrency, speed, and role-specific reasoning
- 06Deployed and demonstrated a working production-grade prototype on Vercel
(Inside)
01
Resume Ingestion & Indexing
Automated resume analysis with private, recruiter-scoped storage and retrieval
02
Candidate Search Interface
Multi-layered discovery using voice or text across skills, projects, and experience
03
Hiring Campaign Builder
JD-based campaign creation with configurable interview depth and question count
04
Dynamic Question Generation
AI-generated, candidate-specific questions spanning fundamentals, systems design, leadership, and standardized culture-fit evaluation
05
AI Interview System
Voice-based interviews via an AI avatar with timestamped response capture
06
Recruiter Analytics Dashboard
Live tracking of interview progress, completion rates, and AI-ranked candidate outcomes
07
Deployment & Demo
Serverless deployment on Vercel, with a recorded demo showcasing the full flow