Case study 14
MUST Hiring & Match Hire
MUST Hiring, built at MUST Company in Seoul, serves both sides of hiring. Candidates build a profile, see an AI match score on every job before they apply, apply in one click with a saved resume and cover letter, and practise with AI-generated interview questions, recording their answers for instant feedback. Recruiters describe a role in plain language to get a full job description in 8+ languages, then work from an ATS dashboard with a searchable candidate database, AI match scores, interview scheduling and hiring analytics. Its successor, Match Hire, is AI-first: a team describes who it wants to hire and gets candidates ranked by fit, with Kyu, an AI hiring assistant, a talent workspace, pipeline automations and a joint inbox.
- Full-stack engineer (MUST Hiring), frontend engineer (Match Hire)
- Enterprise
- May 2024 – Jun 2026
- AI, SaaS
- Recruitment, Applicant Tracking, AI Matching, AI Search, Interview Practice
- Python
- FastAPI
- DynamoDB
- Next.js
- TypeScript
What MUST Hiring does
- AI resume matching, with a match score and recommendations for each job
- AI interview practice with role-specific questions, recorded audio answers and instant feedback
- One-click apply with a saved resume and cover letter
- Guided candidate onboarding, resume management and a career dashboard
- AI job description generator in 8+ languages
- ATS dashboard with KPI cards, funnel charts and top-job insights
- Searchable candidate database with profiles, resumes and match scores
- Interview scheduling linked to applications
- Analytics on hiring performance, recruiter activity and time-to-hire
- Match Hire: deep search that ranks candidates by fit from a plain-language description
- Match Hire: profile cards with CV, persona, skills, trust score and LinkedIn and GitHub enrichments
- Match Hire: Kyu, an AI hiring assistant, plus a talent workspace, automations and a joint inbox
The problem
Candidates apply blind and walk into interviews unprepared, while recruiters spend hours writing job descriptions and sorting applications by hand.
Approach
AI sits at the three points where both sides lose the most time: matching a resume to a job, writing the job description, and preparing for the interview. Everything else is a standard ATS (candidates, applications, interviews and reports) built around those match scores.
Key decisions
- D1
Score before you apply
Candidates see their match on every listing, so applications go where they have a chance and recruiters get better-fitting applicants.
- D2
Practice with your voice
Interview answers are recorded as audio, closer to a real interview than typing.
- D3
Plain language in, JD out
Recruiters describe the role in their own words and get a complete job description in the language they need.
My contributions
- Built the whole product as a full-stack engineer at MUST Company: Python and FastAPI backend, DynamoDB, and a Next.js and TypeScript frontend
- Built the candidate side: profiles, AI match scores, one-click apply and AI interview practice with audio answers
- Built the recruiter side: the AI job description generator, ATS dashboard, candidate database, interview scheduling and analytics
- Worked as a frontend engineer on Match Hire, the AI-first successor, across search, profile cards, the talent workspace, the inbox and reporting