Aptly

Inspiration

Talent is everywhere, but opportunity is not.

That single observation sits at the heart of Aptly. We kept noticing the same frustrating pattern around us: talents from less "prestigious" schools being passed over for internships and jobs not because they lacked ability, but because they lacked the right school crest on their resume or the right person in their network. Hiring decisions were being made on pedigree, not performance.

We wanted to ask: what if the system never saw your school name at all? What if the first thing an employer evaluated was your actual work?

What it does

Aptly is an anonymous talent-matching platform that decouples opportunity from privilege. It serves two sides of the hiring equation:

• Aspiring Opportunists — students, job seekers, and emerging talent who struggle with biased selection and nepotism
• Talent Seekers — employers, mentors, angel investors, and VCs who struggle to find capable individuals beyond paper qualifications

Every candidate goes through the Capability Pipeline — a transparent, multi-stage process designed to surface real talent and eliminate pedigree bias at every step:

1. Build your portfolio — Add projects and work samples. School name is stripped automatically.
2. Complete a challenge — Employers post role-specific tasks with MCQ and written questions tied directly to the job scope.
3. Spoken interview — A voice AI agent conducts a structured interview. Responses are transcribed and scored — employers never see the raw video.
4. Suitability score — Responses are matched against the job description to produce an objective ranking.
5. Fair match — Top-ranked candidates are surfaced to employers with their anonymised portfolio first. The full resume is only revealed after acceptance.

This is the paradigm shift: bias filters out the best — we filter out the bias.

How we built it

Our stack was chosen to build something real and demoable, not just a mockup:

Layer	Technology
Frontend	Next.js
Backend	Node.js
Email	Resend
Database	PostgreSQL
Voice AI Agent	ElevenAgents
Scoring Engine	OpenAI

A hirer creates a posting stored in PostgreSQL. A candidate discovers it, applies, and is routed into an AI-conducted spoken interview powered by ElevenAgents. The voice agent transcribes the session in real time, and the transcript is passed to our OpenAI-powered scoring engine, which evaluates the candidate across four dimensions:

1. Technical Depth
2. Communication
3. Problem Solving
4. Clarity

The hirer receives a ranked list of candidates with anonymised profiles. Only upon acceptance is the full resume unlocked.

Challenges we ran into

Designing true anonymisation without losing signal. Stripping a school name is easy. Stripping implicit signals — graduation years, internship company names, phrasing that reveals background — is much harder. We had to think carefully about what the "Aptly fairness active" layer actually redacts versus what it preserves to keep the profile useful.

Making AI scoring feel fair, not opaque. Candidates and hirers both need to trust the score. We addressed this by producing an executive synthesis alongside the numerical breakdown — a plain-language explanation of what the candidate did well and where they fell short, so the score is never a black box.

Real-time voice interview reliability. Latency in the ElevenAgents voice pipeline was a non-trivial engineering challenge. A stilted, laggy AI interviewer undermines the entire experience. Getting the turn-taking and transcription pipeline smooth enough to feel like a real conversation took significant iteration.

Cold-start on both sides of the marketplace. A two-sided marketplace with zero listings is useless to candidates, and a platform with zero candidates is useless to hirers. We had to think carefully about go-to-market sequencing — seeding the hirer side first with early-access postings to ensure candidates land on a platform with real opportunities from day one.

Accomplishments that we're proud of

We're proud that we have built an MVP in one week. In the time we had, we shipped a real Next.js web app with live postings, a functioning AI voice interviewer, a scoring engine that produces structured feedback, and a two-sided dashboard for both candidates and hirers.

Most of all, we're proud that the demo works end-to-end: a candidate can discover a posting, complete an AI interview, and a hirer can receive a ranked, anonymised result — all in one sitting.

What we learned

Building Aptly taught us that fairness is an engineering problem, not just a values statement. Every design decision — what data to collect, what to display, when to reveal identity — is a lever that either reinforces or dismantles systemic bias. The hardest part wasn't writing the code; it was resisting the temptation to add "just one more field" that would subtly reintroduce the pedigree signals we were trying to eliminate.

We also learned that AI assessment is most powerful when it's transparent. A score with an explanation is a tool. A score without one is just another black box replacing the gut-feel bias we set out to fix.

What's next for Aptly

• Expanding opportunity types — beyond jobs and internships to mentorships, project collaborations, and startup co-founder matching
• Bias audit tooling — giving hirers a dashboard that surfaces whether their acceptance patterns are drifting back toward credential-based selection
• Richer assessment formats — adding take-home coding challenges, design tasks, and case studies alongside the spoken interview
• Institutional partnerships — working directly with polytechnics and universities to give their students a fair shot at opportunities they're currently locked out of
• Mobile app — bringing the full Aptly experience to a native iOS and Android app so candidates can interview and apply from anywhere

The mission stays the same: opportunities for talent, not connections.

Built With

• elevenlabs
• next.js
• node.js
• openai
• postgresql
• typescript