Nestrovar, an AI-powered hiring pipeline for African tech talent.
Nestrovar matches vetted African tech talent with global companies. Skilled African developers stay under-discovered while global hiring stays slow, noisy, and biased. Nestrovar exists to close that gap with AI-driven screening and a hiring workflow that actually moves.
I led design with full ownership across three surfaces: the marketing website, the talent-side product, and the admin console the team uses to run jobs end to end. The admin team currently posts and runs hiring on behalf of employers, with an employer-facing surface as the next major surface in the roadmap.
Spreadsheets, Google Forms, and good people getting lost.
Hiring at the parent company was fragmented across Google Forms, spreadsheets, manual CV reviews, and scattered comms. Recruiters were drowning, candidates were re-entering the same details for every role, and strong people were getting lost in the pile.
AI without trust collapses back to spreadsheets.
The brief was to replace all of that with a single intelligent system. But intelligent systems carry their own design problem: when AI is doing the heavy lifting on screening and shortlisting, how do you make recruiters trust it without making them feel like they’ve handed their judgment over to a black box?
Designing a pipeline, not a stack of screens.
Most weak hiring tools are a stack of disconnected screens that feel fine in isolation but together feel like a maze. I approached Nestrovar as a pipeline instead. Job creation flows into applications, applications into AI shortlisting, shortlisting into interviews, interviews into feedback, feedback into offers. The admin dashboard mirrors that flow so a recruiter always knows where they are, what’s next, and what decision they actually need to make. Cognitive load drops when the product reflects how the work happens.
Readable AI, not magical AI.
The riskiest part of the product was the AI shortlisting layer. If recruiters don’t trust the recommendations, they fall back to manual review and we’ve built nothing. So I designed the AI outputs to be readable, not magical. Every shortlisted candidate gets a structured profiling report – skill scores with years of experience and proficiency, strengths and areas-for-development, a plain-language profile summary, metrics on professional relevance and project impact. Every score is anchored to evidence pulled from the candidate’s work history.
Automation with override.
The AI shortens hours of CV reading to minutes, but it never removes the recruiter from the loop. They can pull the full shortlist, override the ranking, and dig into any profile. Automation with human control – that’s the move that turns a feared tool into a trusted one.
Upload once, use everywhere.
In the old system, every application started from scratch. I built the talent side around one principle: upload once, use everywhere. A candidate creates a structured profile that flows into every application they make. On the admin side, that profile becomes searchable and filterable, so a candidate who didn’t get this role can be rediscovered for the next one. Every application strengthens the pool; every search across the pool gets richer over time.
The branded CV is where the product meets the buyer.
When a recruiter shortlists a candidate, they need something to send to the employer. The old approach meant attaching whatever CV the candidate originally uploaded – different fonts, lengths, polish, all stitched into one inconsistent shortlist. I designed a branded CV template that auto-generates from each candidate’s profile. Same structure, hierarchy, identity, every time. An employer reviewing five shortlisted candidates sees five professional, comparable documents instead of five wildly different files.
Vetted and focused, not open and noisy.
The public site explains who Nestrovar is for and why African tech talent and global companies should both want in. The positioning is intentional: an AI-driven, vetted, focused pipeline, not an open marketplace.
Career tool, not just a job board.
A structured signup and profile flow that captures everything once and reuses it everywhere. A personalized learning roadmap generated from the candidate’s profile, surfacing prioritized areas for skill development so the platform becomes a career tool, not just a job board. Application status visibility at every stage and automated communication so no one is left wondering whether they were rejected or just forgotten.
Where the bulk of the design work lives.
The admin console handles the full hiring lifecycle – roles with custom screening questions and budget guidance against industry benchmarks, a searchable database of every past applicant, AI shortlisting sized to the hire, scheduling with AI-generated interview guides, structured feedback that captures interviewer insights in a comparable way, digital offer creation and tracking, and real-time analytics on pipeline health.
- 01Job management – custom questions, status, budget guidance
- 02Candidate database – searchable, filterable, rediscoverable
- 03AI shortlisting – profiling reports + side-by-side comparison
- 04Interview management – scheduling + AI interview guides
- 05Structured feedback – consistent framework, not free-text rambling
- 06Offer management – digital creation, tracking, response
- 07Analytics – time-to-shortlist, time-to-hire, stage conversion
A pipeline the team can run end to end.
What’s live is a marketing website that positions the product clearly, a talent-side experience designed around reusable profiles and personalized growth, and an admin console that turns a fragmented manual hiring process into an AI-assisted pipeline the team can run end to end. The employer portal is the next major surface in the roadmap.
The stretch was designing for trust, not interfaces.
The project stretched me hardest around designing for trust. Every interaction on the AI shortlisting layer had to earn the recruiter’s confidence before the automation paid off. Get that wrong and the whole system collapses back to spreadsheets. Get it right and you compress hours of CV review into minutes without anyone feeling like they handed their judgment over to a machine.
