convez.ai
PluginProdutividadeIntermediário
TJ-Neary

ResumePRO

TJ-Neary/ResumePRO

IA que entrevista você e gera currículos estratégicos adaptados a cada vaga

Ver no GitHub
InicialPluginScore · de 0 a 100

Ecossistema de busca de carreira alimentado por IA que conduz entrevistas conversacionais STAR para extrair experiências profissionais, analisa ofertas de emprego por múltiplos frameworks e gera currículos otimizados para ATS com bullets verificados e estratégias de posicionamento personalizadas.

Instalação

$npx convez-ai TJ-Neary/ResumePRO

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 9 meses
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Gerar currículos personalizados e otimizados para ATS baseado em análise de vagas específicas
  • Conduzir entrevistas conversacionais STAR para estruturar e validar experiências profissionais
  • Analisar ofertas de emprego através de múltiplos frameworks (MoSCoW, Iceberg Model) para identificar requisitos ocultos
  • Facilitar transições de carreira (incluindo militar-civil) com reposicionamento estratégico de experiências

Veredito

Sistema sofisticado que diferencia-se por usar análise semântica real em vez de matching simples de keywords, com pipeline de duas fases que garante consistência entre estratégia e execução. Excelente para profissionais em transição, mas código-fonte não é público—é mais um case de arquitetura que plugin pronto para deploy imediato.

README

A tradução para o português deste README ainda está na fila.

ResumePRO

Showcase Python Anthropic Tests License

ResumePRO

A comprehensive, AI-empowered career search ecosystem that builds genuine contextual understanding of a user's professional experience through conversational AI interviews, then produces precisely targeted, ATS-optimized resumes matched to specific job postings.

This is not keyword insertion. ResumePRO conducts extensive AI-led interview sessions to construct a semantic model of a user's career history, then applies multi-framework job analysis and evidence-verified experience matching to generate resumes that authentically represent qualifications aligned with each role's requirements. Built for career transitions, including military-to-civilian, across 16 professional role families.

This is a showcase repository. It demonstrates the system architecture, design decisions, and capabilities of a private project. Source code is not included. For technical inquiries, open an issue.


Architecture Highlights

1. Two-Phase Strategic Resume Pipeline

The flagship capability: a two-phase LLM-driven pipeline that acts as a master career coach. Phase A (Strategy Development) analyzes the full experience library, job requirements, and role family profile to produce a reviewable positioning strategy: narrative arc, experience selection with rationale, per-section bullet directives, keyword distribution, and gap mitigation plans. Phase B (Targeted Bullet Writing) takes the approved strategy and writes custom bullets grounded in real STAR data, with a verification gate ensuring every claim traces to documented evidence. The mechanical pipeline (keyword-based scoring and pre-generated bullet selection) remains as an offline fallback.

2. Semantic Experience Understanding

The system goes beyond keyword matching by constructing a rich semantic representation of each experience. Embedding-based similarity scoring compares job requirements against verified professional experiences across six dimensions (keyword match, competency alignment, recency, metric strength, semantic similarity, and personalized assessment signals), while a grounding verification layer ensures every claim traces to documented evidence with explicit confidence levels (Verified, Calculated, Estimated).

3. Conversational STAR Interview Engine

A multi-turn interview orchestrator guides users through structured experience extraction using the STAR methodology (Situation, Task, Action, Result). The system manages interview phases, validates completeness, links evidence artifacts, and stores sessions with full conversation history. Each interview produces structured, reusable experience entries that feed the generation pipeline.

4. Multi-Framework Job Analysis

Job postings are analyzed through three complementary frameworks simultaneously: MoSCoW prioritization (Must/Should/Could/Won't requirements), the Iceberg Model (visible requirements vs. hidden organizational needs), and semantic clustering (grouping related requirements into competency themes). This layered analysis produces a nuanced understanding that drives both experience matching and content generation.


Feature Overview

FeatureTechnical ApproachBusiness Value
Strategic Resume PipelineTwo-phase LLM generation: positioning strategy then targeted bullet writing with grounding verificationResumes read like they were written by a career strategist, not assembled by a keyword matcher
AI-Led STAR InterviewsMulti-turn LLM orchestration with phase management and evidence linkingCaptures rich, verified experiences through natural conversation
Semantic Experience MatchingEmbedding-based similarity with multi-dimensional scoringSurfaces the most relevant experiences for each job posting
Job Posting AnalysisMoSCoW, Iceberg Model, and semantic clustering frameworksUnderstands both explicit requirements and implicit expectations
Evidence VerificationClaim grounding against source documents with confidence labelingEvery metric traces to evidence: no fabricated numbers
ATS OptimizationClean DOCX export with keyword-aware formattingResumes parse correctly through applicant tracking systems
Cover Letter GenerationLLM-powered with pain-point strategy engineAligned messaging across resume and cover letter
Interview PreparationSTAR story selection matched to likely interview questionsConsistent narrative from resume through interview
Company ResearchAutomated company profiling with culture and values analysisInformed positioning for each application
Human-in-the-Loop EditingBullet swap, remove, reorder, undo history, draft versioningFull control over generated content before export
Assessment IntegrationProfessional assessment data as a personalized scoring signalRanks experiences by alignment with individual strengths
Data StandardizationAutomatic schema normalization across all experience entriesConsistent data quality regardless of input format
16 Role FamiliesConfigurable role profiles with family-specific bullet librariesCovers HR, PM, OD, L&D, Change Management, AI Transformation, and more

Screenshots

Screenshots and demo recordings coming soon.

Planned visuals: CLI job analysis output, generated resume sample, interview session flow, experience matching dashboard.


By the Numbers

MetricValue
Passing Tests2,656 across 104 test files
Source Modules153 files across 25+ packages
CLI Commands31 (23 command groups + 8 standalone)
Role Families16 configurable career profiles
Experience Library36 verified STAR-format entries
Bullet Files42 role-family bullet libraries
Generated Bullets~2,474 role-specific resume bullets
Scoring Dimensions6 (keyword, competency, recency, metrics, semantic, assessment)
Prompt Templates18 YAML-based prompt templates
Development EPICs23 (233+ backlog items)
Resume GenerationTargeted resume in under 60 seconds

Technology Stack

LayerTechnologyRole
LanguagePython 3.12+Core implementation
LLM ProviderAnthropic Claude (Sonnet/Opus)Content generation, strategic positioning, analysis, interviews
LLM FallbackGoogle GeminiAlternative provider support
CLI FrameworkTyperCommand-line interface
Data ValidationPydantic v2Schema enforcement and serialization
Embeddingssentence-transformers (all-MiniLM-L6-v2)Semantic similarity computation
Vector StorageSQLiteEmbedding persistence and retrieval
Document Exportpython-docxATS-optimized DOCX generation
Prompt ManagementYAML templates with cachingStructured prompt loading and versioning
TestingpytestUnit, integration, and regression testing
NLPspaCy, NLTKText processing and entity extraction

System Design

For the complete architecture including C4 diagrams, data flow, and design decision rationale:

View Full Architecture Documentation


About

ResumePRO was designed to solve a real problem: career transitions require deeply personalized resumes that authentically represent transferable experience. Traditional resume tools rely on keyword stuffing or generic templates. ResumePRO takes a fundamentally different approach by first understanding the user's experience through structured interviews, then applying that understanding to each specific job opportunity.

The two-phase strategic pipeline represents the evolution from mechanical resume assembly to genuine career coaching: the system develops a positioning strategy that reads between the lines of job postings, then writes custom content grounded in verified experience data.


Copyright 2026 TJ Neary. All Rights Reserved.

Ver o original no GitHub

Bom para

Relacionados