Back to projects

Navero

AI-powered hiring platform automating candidate evaluation with video screening, transcription, and workflow automation.

Full-Stack Engineer|May 2025 - Present|Navero
Next.jsNestJSTypeScriptPythonGCPSupabaseAssemblyAI
98% P99 latency reduction
18% activation rate improvement
20+ analytics events tracked

Navero — AI-Powered Hiring Platform

The Problem

Hiring managers spend hours manually reviewing candidate video screenings, taking notes during interviews, and moving candidates through ATS stages. The process is slow, inconsistent, and prone to human bias.

The Solution

I helped build Navero, an AI-powered hiring platform that automates the entire candidate evaluation pipeline — from video screening to interview transcription to automated scoring and stage transitions.

What I Built

Multi-Region Architecture (V2 Migration)

Led the V2 architecture migration with data residency and regional isolation to comply with GDPR, KSA, and UAE market requirements. Each region gets its own data partition while sharing the application layer.

Event-Driven Video Screening

Designed an async pipeline using GCS Cloud Functions and AssemblyAI transcription webhooks. Candidates record video responses; the system automatically transcribes, analyzes, and scores them without human intervention.

Hiring Workflow Engine

Architected a modular workflow engine supporting Assessment, Screening, and Intake stages. Hiring managers define custom evaluation criteria, and candidates are automatically scored and transitioned across pipeline stages using rule-based automation.

Performance Optimization

Reduced the dashboard P99 latency from 60 seconds to 1 second — a 98% improvement. This involved redesigning data access patterns, implementing efficient filtering strategies, and optimizing database queries.

Analytics & Monitoring

Implemented Mixpanel analytics across 20+ key events, enabling identification of user drop-off points and improving user activation rate by 18%. Integrated Sentry for production monitoring across all Cloud Run services.

Tech Stack

  • Frontend: Next.js, TypeScript, Tailwind CSS
  • Backend: NestJS, Python
  • Cloud: GCP (Cloud Run, Cloud Functions, Cloud Tasks, GCS)
  • Database: Supabase (PostgreSQL)
  • AI/ML: AssemblyAI, RecallAI
  • Monitoring: Sentry, Mixpanel

Key Results

  • 98% P99 latency reduction (60s → 1s)
  • 18% improvement in user activation rate
  • Multi-region architecture for GDPR/KSA/UAE compliance
  • Zero manual intervention in video screening pipeline