VINCONS Recruitment CRM

AI-assisted recruitment operations for construction teams, unifying candidate intake, data validation, interviews, employee codes, and reporting in one secure workflow.

Comment

Inspiration

What it does

How we built it## Inspiration

Construction recruitment teams often receive candidate information through phone calls, social media messages, handwritten records, and spreadsheets. The same information must then be checked and re-entered by several employees, creating delays, inconsistent records, and privacy risks.

VINCONS Recruitment CRM was created to provide one consistent operational workflow for recruitment teams managing construction workers at scale.

What it does

VINCONS Recruitment CRM connects the main stages of recruitment in one desktop-first system:

  • Candidate intake from manual entry and supported recruitment channels
  • Interview and application tracking
  • Vietnamese name, phone number, address, bank, trade, and salary normalization
  • Identity and recruitment-record validation
  • Employee-code assignment
  • Cross-stage candidate synchronization
  • Excel import and export while preserving the required workbook structure
  • Recruitment dashboards and consolidated reporting
  • Local encryption for sensitive identity-document assets
  • Configurable AI-assisted parsing and OCR workflows

The product helps recruitment staff transform incomplete or inconsistent candidate information into structured operational records.

How we built it

The desktop application is built with Electron, Node.js, JavaScript, HTML, and CSS. Shared modules provide parsing, fuzzy matching, validation, local encryption, Excel processing, provider integrations, and synchronization across the recruitment workflow.

The repository also includes a recruitment website and an optional SaaS architecture for future deployment.

The project existed before OpenAI Build Week. During the submission period, we used Codex and GPT-5.6 to improve the data-normalization workflow, review implementation and privacy risks, expand automated verification, and prepare a reproducible judging repository.

The current self-test suite passes 81 checks covering:

  • Vietnamese administrative catalogs
  • Salary and trade normalization
  • Phone-number and bank-account parsing
  • Local encryption and decryption
  • Excel round-trip processing
  • Candidate validation
  • Address conversion and fuzzy matching
  • OCR result normalization
  • Synchronization between recruitment stages

Challenges we ran into

One major challenge was distinguishing similar numeric fields such as phone numbers, bank accounts, identity numbers, and salary values when candidates sent unstructured messages.

Vietnamese names, administrative addresses, trade names, and bank names may also be written in many different forms. We combined deterministic normalization, fuzzy matching, validation rules, and configurable AI assistance to improve consistency.

Another challenge was preserving existing Excel workbook structures while safely inserting recruitment records. Privacy was equally important because production candidate data may contain identity documents and other personal information.

Accomplishments that we are proud of

  • Built a coherent workflow spanning candidate intake, interviews, data entry, employee-code assignment, and reporting
  • Created reusable Vietnamese recruitment-data normalization modules
  • Added local encryption for sensitive identity-document assets
  • Preserved workbook structures during Excel import and export
  • Created an automated test suite with 81 passing checks
  • Prepared a privacy-safe judging repository without production candidate records, credentials, tokens, or private encryption keys

What we learned

We learned that reliable recruitment automation requires more than extracting text. The system must understand the operational meaning of each field, preserve existing business processes, protect personal information, and make errors visible before records move to the next stage.

Codex was especially useful for reviewing interactions across modules, identifying edge cases, expanding verification, and documenting a reproducible setup for judges.

What's next for VINCONS Recruitment CRM

Next steps include:

  • A hosted sandbox containing synthetic candidate data
  • More granular role-based permissions
  • Stronger audit logging and approval workflows
  • Expanded GPT-5.6-assisted extraction and OCR evaluation
  • Additional recruitment-channel integrations
  • Analytics for recruitment cost, conversion, and workforce demand

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for VINCONS Recruitment CRM

Built With

Share this project:

Updates

Submission history