Certum AI
Voice‑powered mock interviews using Hume AI and Gemini for performance feedback.
Visit Live Site↗
An innovative interview preparation platform that leverages AI to simulate realistic job interviews and provide actionable feedback on performance.
Overview
Certum AI is a cutting-edge interview preparation platform that leverages empathic voice technology and emotion analysis to transform how job seekers practice. Built with Next.js 15 and React 19, the application simulates realistic interview environments through an emotion-aware AI interviewer. By combining Hume AI’s voice interface with Google Gemini’s reasoning capabilities, Certum AI provides real-time feedback on both technical accuracy and emotional delivery—addressing not just what a candidate says, but how they say it.
Experience Certum AI↗My Role
As the Lead AI Architect, I engineered a multi-model system that synchronizes Hume AI’s empathic voice technology with Google Gemini’s reasoning to provide real-time emotion analysis and technical feedback. I architected the full-stack infrastructure using Next.js 15 and Drizzle ORM, implementing a robust security layer with Arcjet for rate limiting and Clerk for role-based access control. My work focused on optimizing the user experience through streaming AI responses and a granular caching strategy that ensures high-performance data retrieval across complex interview workflows.
A visual breakdown of the core platform capabilities, including voice-based mock interviews, AI resume screening, and real-time performance tracking.
Key Features
Certum AI delivers an immersive interview experience through an empathic voice system that tracks emotional cues like confidence and anxiety in real-time. The platform offers context-aware question generation tailored to specific job descriptions and experience levels, complemented by a comprehensive resume analyzer that scores ATS compatibility and keyword coverage. To ensure data-driven growth, users receive structured AI feedback and complete session transcripts, all protected by a robust security layer and a credits-based permission system.
Dynamic social proof showcasing high user success rates, thousands of interviews conducted, and significant improvements in candidate confidence.
Key Takeaways
Building Certum AI allowed me to master the orchestration of Streaming AI responses, utilizing the Vercel AI SDK to provide users with instantaneous, conversational feedback. I tackled the challenge of "Emotion-Data Integration"—learning how to capture high-dimensional emotional features from a voice stream and translate them into actionable career advice. Additionally, the project sharpened my skills in PostgreSQL schema design, specifically in creating cascading data models that ensure absolute data integrity in a multi-user SaaS environment.
AI-generated actionable feedback providing specific structural improvements and keyword optimizations to enhance ATS compatibility.
Certum AI represents the next generation of career coaching, where AI moves beyond text to understand the nuances of human speech and emotion. By architecting a platform that bridges the gap between technical preparation and emotional intelligence, I created a tool that provides users with a safe, high-stakes environment to refine their professional presence. This project serves as a definitive showcase of my ability to handle complex API integrations, secure serverless architectures, and modern UI/UX principles.
© 2026 NEMANJA