Screnly — AI Interview Evaluation Platform for Businesses

A B2B hiring tool that lets companies run fully automated AI interviews. Businesses post a job, candidates apply and get interviewed by an AI in real time, and a detailed evaluation report is automatically generated and sent — so hiring teams can focus on shortlisting, not screening.

The problem it solves

First-round screening is slow, expensive, and inconsistent. Recruiters spend hours interviewing candidates who aren't a fit, while great candidates get lost in the pile. Screnly automates the entire first-round interview — saving time on both sides and giving every candidate a fair, structured evaluation.

How it works

1. Business posts a job — with a full job description inside the platform

2. Candidate receives an invite link — and submits their CV to apply

3. Real-time AI interview happens — the AI reads both the CV and the job description to ask relevant, personalized questions via a live video avatar

4. Evaluation report is generated — scoring the candidate's answers, highlighting strengths and gaps

5. Report is emailed automatically — to the business owner right after the interview ends

6. Business shortlists from dashboard — with all candidates and reports in one place

What I built

Business dashboard and job management

Companies create job listings with a full description and share an invite link with candidates. All applicants, their interview status, and evaluation reports are managed from a central dashboard — including a shortlisting workflow for hiring decisions.

CV submission and candidate onboarding

Candidates apply by submitting their CV through the platform. The AI uses both the uploaded CV and the job description to generate a fully personalized interview — no generic questions, no manual prep required by the business.

Real-time AI video interview

Powered by HeyGen's streaming avatar API and OpenAI Realtime API, the AI conducts the interview live — asking relevant follow-up questions, listening to voice answers, and adapting in real time. ElevenLabs provides natural, human-sounding speech so the experience feels genuine.

Automated evaluation and report generation

After each interview session ends, the platform automatically generates a detailed candidate evaluation — scoring answers, summarizing communication quality, and flagging key strengths or gaps relevant to the job description.

Email delivery pipeline

The report generation and email sending pipeline runs fully serverlessly via Supabase Edge Functions — triggered automatically the moment a session ends, with no manual action required from the business.

Tech stack

• Frontend: React

• Backend: Supabase Edge Functions (serverless)

• Database: PostgreSQL via Supabase

• Auth: Supabase Auth

• AI interview: OpenAI Realtime API

• Video avatar: HeyGen Streaming Avatar API

• Voice: ElevenLabs Text-to-Speech

• Email delivery: Automated via Edge Functions

Key accomplishments

• Built a complete end-to-end hiring pipeline — from job post to candidate shortlist — in a single platform

• Integrated 5 AI APIs into one seamless, real-time flow with under 1 second voice-to-response latency

• Eliminated manual first-round interviews for businesses entirely

• Automated evaluation report generation and email delivery — zero manual steps after interview ends

• Designed and shipped the entire product solo: architecture, design, development, and deployment

What I learned

• Synchronizing real-time voice, streaming video avatars, and WebSocket events without breaking the illusion takes precise sequencing — a small timing mistake immediately makes the experience feel fake

• Building a product that feels human requires as much UX thinking as engineering — the technical stack is only half the challenge

• Supabase Edge Functions handle scalable serverless workflows well but require careful design around cold starts and execution timeouts

• CV parsing combined with job description context produces dramatically better interview questions than a generic prompt

Why this matters

Screnly replaces the most time-consuming part of hiring — the first-round interview. It gives businesses a scalable, consistent way to evaluate candidates at any volume, and gives candidates a fair shot regardless of scheduling constraints. This is the kind of full-stack AI product work I bring to every project.