One intelligent pipeline from job creation to offer decision
80%
Reduction in administrative hiring overhead through end-to-end automation
Sub-24h
Average time-to-first-interview from job creation to candidate scheduling
3x
Improvement in evaluation consistency with AI-assisted multi-modal scoring
Recruitment was fragmented, slow, and biased
Enterprise hiring teams relied on a patchwork of disconnected tools — separate platforms for job boards, scheduling, video interviews, scoring rubrics, and applicant tracking. This fragmentation created data silos, forced manual data entry between systems, and introduced unconscious bias through inconsistent evaluation criteria. For high-volume e-commerce and retail organizations, the bottleneck was especially acute. HR managers spent 60-70% of their time on administrative overhead: manually reviewing CVs, scheduling interviews across time zones, transcribing interview recordings, and consolidating feedback from multiple panelists.
A unified, AI-native hiring operating system
Selectra HR was architected as a modular, full-stack platform built on Next.js 15 and a Python FastAPI backend, orchestrating a constellation of AI services into a single cohesive workflow. From the moment an HR manager creates a job, AI agents begin working in parallel: OpenRouter-powered recommendation engines suggest role-specific culture-fit games and interview questions; Groq LLaMA-3 drives ultra-low-latency real-time voice interviews with adaptive follow-ups; and automated scoring pipelines evaluate voice responses, video posture, and human panelist inputs into a unified composite score.
Core Capabilities
AI Voice Interview Agent
Groq-powered real-time conversational interviews that use LLaMA-3 for ultra-low-latency responses. The agent adapts questions based on candidate answers, supports bilingual Urdu/English modes, and features configurable accents for a natural, localized experience.
Culture-Fit Gamification Engine
A library of 30+ gamified assessments (Priority Tower, Brand Storyteller, Crisis Commander) that measure emotional intelligence, risk tolerance, collaboration, and role-specific traits — dynamically selected by AI based on job description, department, and experience level
Multi-Modal Scoring & Evaluation
A trifecta of scoring inputs: AI voice analysis that grades answer accuracy and completeness, computer vision-based posture analysis from Jitsi recordings, and structured human-panelist scoring — all merged into a unified composite score with hiring recommendation
Automated Interview Scheduling
Google Calendar API integration for creating Meet links, automated SMTP email notifications to candidates and panelists, and a 60-second background scheduler loop that automatically assigns candidates to slots
Real-Time Analytics Dashboard
Interactive Recharts-powered visualizations including hiring funnel conversion rates, score distribution histograms, application volume trends by department, and live candidate pipeline tracking
n8n-Driven LinkedIn Integration
Automated workflow that publishes approved job postings to LinkedIn directly from the platform, coupled with a custom companion script that simplifies the n8n posting webhook for non-technical HR operators
Architecture & Technology
Frontend
- Next.js 15 (App Router)
- React 18 with TypeScript
- Tailwind CSS 4
- Radix UI Primitives
- Recharts
- React Hook Form + Zod
- Lucide React Icons
- Supabase JS Client
Backend
- Python 3.11 / FastAPI
- Supabase (PostgreSQL)
- OpenRouter AI (Gemini/Groq models)
- Groq LLaMA-3 / Mixtral
- OpenAI Whisper (STT)
- Groq TTS (Text-to-Speech)
- Google Calendar API
- SMTP Email Service
Infrastructure
- Hetzner CX32 VPS (8GB RAM
- 4 vCPU)
- Docker & Docker Compose
- Vercel (Frontend Hosting)
- Jitsi Meet (Self-Hosted Video)
- Backblaze B2 (S3-Compatible Storage)
- n8n Workflow Automation
External APIs
- OpenRouter API
- Groq API
- Supabase API
- Backblaze B2 API
- Google Calendar API
- LinkedIn API (via n8n)