A platform that helps candidates prepare better job applications. It reads job descriptions, rewrites resumes to match, generates cover letters, runs mock interviews with feedback, and pulls together research on the company. Focused on the Brazilian market for now, with international expansion as the next step.
A platform that helps candidates prepare better job applications. It reads job descriptions, rewrites resumes to match, generates cover letters, runs mock interviews with feedback, and pulls together research on the company. Focused on the Brazilian market for now, with international expansion as the next step.
I wanted to make a product end-to-end on my own. Design, data science, frontend, backend, AI pipeline. The career space felt right because the job application process feels genuinely broken for job seekers. Most resumes get tossed by ATS systems before a human ever sees them and candidates are not aware.
I wanted to make a product end-to-end on my own. Design, data science, frontend, backend, AI pipeline. The career space felt right because the job application process feels genuinely broken for job seekers. Most resumes get tossed by ATS systems before a human ever sees them and candidates are not aware.
The CV tool market is dominated by dated, form-heavy interfaces. Functional, but undifferentiated. I treated design quality as a competitive advantage , deliberate interactions, smooth transitions, and micro-animations that make the product feel more refined than anything else in the space. This project also changed how I design a large product. Rather than starting in Figma I designed directly in code with AI as a collaborator. The trade-off in upfront fidelity was worth the speed and freedom to experiment and deliver functioning code.
The CV tool market is dominated by dated, form-heavy interfaces. Functional, but undifferentiated. I treated design quality as a competitive advantage , deliberate interactions, smooth transitions, and micro-animations that make the product feel more refined than anything else in the space. This project also changed how I design a large product. Rather than starting in Figma I designed directly in code with AI as a collaborator. The trade-off in upfront fidelity was worth the speed and freedom to experiment and deliver functioning code.
AI Engine
AI Engine
The system runs on a multi-stage pipeline that connects job description analysis, scoring, guided editing, and continuous re-evaluation.
When a user submits a job description, the pipeline extracts structured requirements and maps them against the candidate's CV, producing a compatibility breakdown across five weighted components. The scoring is designed to approximate ATS screening logic, not replicate it exactly, but give candidates a persistent, reliable signal of where they stand. That data flows into the CV editor and the AI chat.
The chat has full context, job requirements, scoring breakdown, current CV state, and can make direct edits to the document on behalf of the user. Every change triggers a re-evaluation: the system triangulates the updated CV against the original requirements, recalculates scores, and surfaces new suggestions. The loop is continuous. Edit, re-score, refine.
The system runs on a multi-stage pipeline that connects job description analysis, scoring, guided editing, and continuous re-evaluation.
When a user submits a job description, the pipeline extracts structured requirements and maps them against the candidate's CV, producing a compatibility breakdown across five weighted components. The scoring is designed to approximate ATS screening logic, not replicate it exactly, but give candidates a persistent, reliable signal of where they stand. That data flows into the CV editor and the AI chat.
The chat has full context, job requirements, scoring breakdown, current CV state, and can make direct edits to the document on behalf of the user. Every change triggers a re-evaluation: the system triangulates the updated CV against the original requirements, recalculates scores, and surfaces new suggestions. The loop is continuous. Edit, re-score, refine.
Now in growth mode, running analytics on user behavior, iterating on the product based on real usage data, and pushing distribution across organic social, paid search, and product-led loops. The kind of work that competitors staff entire teams for, done solo.
Now in growth mode, running analytics on user behavior, iterating on the product based on real usage data, and pushing distribution across organic social, paid search, and product-led loops. The kind of work that competitors staff entire teams for, done solo.
visit prepara.cv