Skip to content
    AndrewBuilds
    All projects

    Career Technology · AI

    Gradr

    An AI-powered career accelerator designed to help students, graduates, and early-career professionals build stronger resumes, prepare for interviews, and manage their career journey.

    Previously known as CareerFlow OS.

    Built with LovableAIReact
    Try the live demo

    Problem

    Early-career candidates get almost no useful feedback. Resumes disappear into applicant tracking systems without explanation, interview practice is either expensive or unstructured, and applications end up scattered across spreadsheets, inboxes and browser tabs. The missing piece is not more advice — it is a system that reviews your material, tells you exactly what is weak, and keeps the whole search in one place.

    What I built

    • AI resume analysis
    • ATS scoring
    • Job tracking
    • AI mock interviews
    • Interview analytics
    • Career development tools

    Built by Andrew

    Inside the product

    The main screens and what each one does.

    • ATS score
      Keyword gaps
      Bullet rewrites
      Section checks

      Resume analysis

      Upload a resume and get a structured breakdown: ATS score, keyword coverage, weak bullet points and concrete rewrite suggestions instead of vague feedback.

    • Role selection
      Live Q&A
      Answer scoring
      Session replay

      AI mock interviews

      Role-specific interview sessions that ask follow-up questions, then score answers on structure, specificity and relevance so practice actually compounds.

    • Saved jobs
      Stages
      Reminders
      Match signals

      Job tracker

      Every application in one board with stage, source and reminders, so nothing is lost between applying and following up.

    • Score trend
      Skill gaps
      Next actions
      Weekly summary

      Career plan & analytics

      A progress view that turns sessions and applications into trends — where scores improve, where they stall, and what to work on next.

    Tech stack

    • Interface

      ReactTypeScriptTailwind CSSComponent library

    • AI layer

      LLM APIsStructured promptingScoring rubricsStreaming responses

    • Backend

      PostgresRow-level securityAuthFile storageServer functions

    • Platform

      LovableMCP integrationCredit-based usage metering

    Challenges & how I solved them

    • Making AI feedback specific, not generic

      The first versions returned polite but useless advice. I moved from open-ended prompts to fixed scoring rubrics with structured output, so every response has to point at a real line in the resume or answer and explain what to change.

    • Approximating ATS behaviour honestly

      Real applicant tracking systems are closed. Instead of pretending to replicate one, the score is built from things that are actually checkable — parseability, section structure, keyword coverage against the job description and formatting risks — and the app explains each component.

    • Keeping AI cost predictable

      Long resumes and interview transcripts get expensive fast. Requests are trimmed and cached, heavy analysis runs on demand rather than on every keystroke, and usage is metered with a credit balance so cost stays bounded per user.

    • Data privacy on personal documents

      Resumes are sensitive. Everything is scoped per user with row-level security, files live in access-controlled storage, and no document is readable across accounts.

    Results so far

    Status
    In active developmentCore flows working end to end.
    Core modules
    6Resume, ATS, interviews, tracker, analytics, plan.
    Feedback loop
    Analyse → practise → applyOne connected system.
    Integrations
    MCP toolsResumes, jobs and reminders exposed to assistants.

    Related work