HIREBASE: A CLOUD-NATIVE PLATFORM FOR AI-ASSISTED RECRUITMENT WITH A VOICE INTERVIEW AGENT

Source abstract

Hiring is still one of the most manual parts of running an organization. Recruiters then manually read through the piles of resumes, follow up with candidates through email for tests and interview appointments, and evaluate first round interviews based on a gut feeling for the right candidate. The tools developed to rectify this one by one are limited in their scope, e.g. a job board, a keyword-matching ATS. The other part of the process remains in spreadsheets and inboxes. To address this need, we've designed and deployed a cloud-native recruitment platform, Hirebase, to close this gap. Publish jobs, score candidates using a large language model (LLM) on resumes, assess candidates with timed performance-based tests and automatically grade, and interview candidates in the first round using a real-time voice AI agent (all within a single system) with recruiters having the final say on all hires. The platform is built with a React/TypeScript front-end, a FastAPI back-end, and Supabase for the PostgreSQL Database and object storage, a dedicated LiveKit voice-agent worker, and an auto-deploying CI/CD pipeline to Vercel, Azure App Service, and Azure Container Instances. Describe the system architecture, engineering decisions made, and results of testing the system end-to-end. We also cover the areas where we believe the platform is lacking and what we believe should be built next! We will not be providing some single algorithm, but instead a working, deployed reference architecture demonstrating how a small team can integrate LLM-based scoring into a real-time speech AI and standard cloud services into one seamless hiring pipeline, without resorting to a multi-agent framework. Keywords: recruitment automation, AI-assisted hiring, resume scoring, voice interview agents, WebRTC, cloud-native systems, decision support.